Abstract
Protean agonists are of great pharmacological interest as their behavior may change in magnitude and direction depending on the constitutive activity of a receptor. Yet, this intriguing phenomenon has been poorly described and understood, due to the lack of stable experimental systems and design strategies. In this study, we overcome both limitations: First, we demonstrate that modulation of the ionic strength in a defined experimental set-up allows for analysis of G protein–coupled receptor activation in the absence and presence of a specific amount of spontaneous receptor activity using the muscarinic M2 acetylcholine receptor as a model. Second, we employ this assay system to show that a dualsteric design principle, that is, molecular probes, carrying two pharmacophores to simultaneously adopt orthosteric and allosteric topography within a G protein–coupled receptor, may represent a novel approach to achieve protean agonism. We pinpoint three molecular requirements within dualsteric compounds that elicit protean agonism at the muscarinic M2 acetylcholine receptor. Using radioligand-binding and functional assays, we posit that dynamic ligand binding may be the mechanism underlying protean agonism of dualsteric ligands. Our findings provide both new mechanistic insights into the still enigmatic phenomenon of protean agonism and a rationale for the design of such compounds for a G protein–coupled receptor.
Introduction
G protein–coupled receptors (GPCRs) comprise the largest superfamily of cell surface receptors, accounting for 3% of all genes in the human genome (Fredriksson and Schiöth, 2005). Upon activation, GPCRs undergo a global conformational rearrangement to transduce their signals from extracellular stimuli into the intracellular environment. Via this mechanism, binding sites for cytosolic adaptor proteins, for instance heterotrimeric G proteins, become exposed (Rasmussen, et al., 2011; Kruse et al., 2013), which then serve to propagate the signaling within the cell (Pierce et al., 2002; Lefkowitz, 2004). GPCRs are versatile membrane proteins that exist in ensembles of possible conformations (Manglik and Kobilka, 2014), and ligands can be classified according to their ability to stabilize subsets of these conformations (Rosenbaum et al., 2009). For instance, agonists favor active receptor conformations over inactive states. Of note, receptors can also spontaneously transit from inactive to active states and vice versa (Costa and Herz, 1989; Leff, 1995).
In addition to the binding site for the endogenous messenger, GPCRs carry distinct druggable sites, which are designated as allosteric sites. Compounds binding to these allosteric binding sites may alter the efficacy and/or binding affinity of orthosteric ligands but may also exhibit intrinsic efficacy for receptor activation/inactivation in their own right (Christopoulos, 2002). Allosteric modulation has extensively been studied at muscarinic acetylcholine receptors (mAChRs), which are prototypal class A (rhodopsin-like) GPCRs. Recent studies have also described dualsteric ligands, which can simultaneously occupy the orthosteric and an allosteric binding site of the M2 subtype of mAChRs. This results in a unique receptor-binding profile and signaling pattern (Steinfeld et al., 2007; Gregory et al., 2010; Bock et al., 2012; Schrage and Kostenis, 2016). Dualsteric/Bitopic ligands can switch between two binding poses, that is, either a dualsteric or a purely allosteric binding pose (Bock et al., 2014, 2016) and thus may stabilize not only one but at least two distinct subsets of receptor conformations (dynamic ligand binding) (Bock et al., 2014).
A particularly fascinating, albeit seldom described, class of GPCR ligands are protean agonists, that is, compounds that display agonism in quiescent receptor systems with low levels of spontaneous receptor activity and inverse agonism in constitutively active systems (Kenakin, 1995). One explanation for this phenomenon may be that protean agonists stabilize a receptor conformation with lower efficacy for a certain signaling pathway than the spontaneously active conformation of a given GPCR. Thus, even though the intrinsic efficacy of the ligand does not vary, its effect changes with the amount of spontaneously active receptors that are present within the experimental system (Kenakin, 1995, 1997).
Up to now, protean agonists have only been identified for a handful of receptors, for instance cannabinoid receptor 2 (Yao et al., 2006; Mancini et al., 2009; Xu et al., 2010; Bolognini et al., 2012), histamine receptor 3 (Gbahou et al., 2003), α-2A adrenergic receptor (Jansson et al., 1998; Pauwels et al., 2002), serotonin receptor 1B (Newman-Tancredi et al., 2003), bradykinin receptor 2 (Fathy et al., 1999; Marie et al., 1999), and β-2 adrenergic receptor (Chidiac et al., 1994, 1996). Nevertheless, protean agonists are highly valuable tools, as they might be most useful for the identification of ligand-specific active states of GPCRs (Kenakin, 2001). Furthermore, protean agonists may represent a promising new class of pharmacologically active compounds in targeted drug therapy for those cases in which receptor mutations cause constitutive activity (Tao, 2008). Under these conditions, protean agonists can re-establish the original tone of receptor activation (Jansson et al., 1998; Mancini et al., 2009). Moreover, the classification of ligands as protean agonists is of significant importance and not trouble-free, because they may be mistaken for inverse agonists in in vitro assays with high levels of constitutive activity (Chidiac et al., 1994), but act as agonists in vivo (Yao et al., 2006; Mancini et al., 2009; Xu et al., 2010).
The rather few examples of GPCR protean agonists are, at least in part, due to the lack of experimental systems in which the amount of spontaneous GPCR activity is reliably and reproducibly controlled. Moreover, specific design strategies to generate protean ligands are missing. In this study, we overcome both limitations: we have developed an assay system to enable such control and employ it in this study to demonstrate that the dualsteric design approach may serve as a novel means to rationally create protean agonists using muscarinic acetylcholine receptor 2 (M2AChR) as a model system.
In particular, we establish and validate experimental conditions that permit analysis of M2 receptor–mediated Gi protein activation, in the absence and presence of a defined amount of spontaneous receptor activity. We found the following: First, dualsteric ligands were able to induce protean agonism at M2AChR. Second, we identified specific pharmacophoric elements that have to be combined to induce protean agonism at this receptor subtype. Third, we propose that dynamic ligand binding (Bock et al., 2014, 2016) may be the underlying molecular mechanism of protean agonism of dualsteric compounds targeting M2AChRs.
Therefore, our findings deliver a new experimental system to study spontaneous activity of GPCRs, two new protean agonists for M2AChR, and a new strategy along with a potential molecular mechanism to achieve protean agonism at this receptor subtype, which may pinpoint to the design of protean agonists at other GPCRs.
Materials and Methods
Materials and Reagents.
Acetylcholine (ACh) iodide, atropine sulfate, and pilocarpine hydrochloride were obtained from Sigma-Aldrich Chemie (Steinheim, Germany). Guanosine 5′-O-(γ-[35S]thio)triphosphate ([35S]GTPγS) and [3H]N-methylscopolamine bromide ([3H]NMS) were from PerkinElmer Life and Analytical Sciences (Homburg, Germany). Cell culture media, HEPES, N-methyl-d-glucamine, sodium bromide, bovine serum albumin (BSA), GDP, and dl-dithiothreitol (DTT) were purchased from Sigma-Aldrich (Taufkirchen, Germany). Magnesium chloride (MgCl) and sodium chloride (NaCl) were acquired from Acros Organics (Geel, Belgium). Na2EDTA was purchased from Applichem (Darmstadt, Germany). Tris(hydroxymethyl)-aminomethane (Tris) and potassium chloride (KCl) were acquired from Carl Roth (Karlsruhe, Germany).
Chemical Synthesis.
The synthesis of all compounds that are not commercially available has been performed following the protocols described elsewhere: iperoxo (iper) (Klöckner et al., 2010); isox and oxo-oxotremorine M (OOM) (Dallanoce et al., 1999); isox-6-naphthalimide (naph) (Disingrini et al., 2006); isox-8-naph and 8-naph (Bock et al., 2014); iper-6-naph, 6-naph, and 6-phthalimide (phth) (Antony et al., 2009); iper-8-naph (Bock et al., 2012); isox-6-phth (Disingrini et al., 2006); iper-6-phth (Antony et al., 2009); and OOM-6-naph and OOM-6-phth (Disingrini et al., 2006). All derivatives were obtained with comparable yields and the same analytical purity as reported in the literature.
Cell Culture.
Flp-In-Chinese hamster ovary (CHO) cells stably expressing the human muscarinic acetylcholine M2 receptor (CHO-hM2 cells) were cultured in Ham’s nutrient mixture F-12 (Ham’s F-12) supplemented with 10% (v/v) fetal calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine at 37°C in a 5% CO2 humidified incubator. The cells were passaged by trypsinization at nearly confluence.
Membrane Preparation.
CHO-hM2 cells were grown to 90% confluence and treated with fresh medium supplemented with 5 mM sodium butyrate for 18–20 hours. On the day of the membrane preparation, medium was aspirated and 2.4 ml ice-cold harvesting buffer (20 mM HEPES, 10 mM Na2EDTA, pH 7.4) was added. Cells were mechanically detached with a cell scraper (Sarstedt AG, Nümbrecht, Germany) before the cell suspension was homogenized using a Polytron homogenizer (1 × 25 seconds and 1 × 20 seconds, level 6). The suspension of lysed cells was centrifuged (10 minutes, 40,000 × g, 2°C), and the pellet was resuspended in storage buffer (20 mM HEPES, 0.1 mM Na2EDTA, pH 7.4). This centrifugation step was repeated twice. The remaining pellet was resuspended in an adequate amount of [35S]GTPγS assay buffer (50 mM Tris, 2 mM MgCl, 1 mM Na2EDTA, 1 mM DTT, pH 7.4) and stored at −80°C (Schrage et al., 2013).
Equilibrium-Binding Assays.
To estimate apparent agonist-binding affinities, membranes from CHO-hM2 cells were diluted in assay buffer, that is, Tris buffer (50 mM Tris, 2 mM MgCl, 1 mM Na2EDTA, 1 mM DTT, pH 7.4) or Tris NaCl buffer (50 mM Tris, 2 mM MgCl, 1 mM Na2EDTA, 1 mM DTT, 200 mM NaCl pH 7.4), supplemented with 100 µM GDP. A total of 30 µg/ml membrane suspensions was then incubated with 0.2 nM [3H]NMS and different concentrations of test compound in a 96-well microtiterplate (Thermo Scientific ABgene, Schwerte, Germany) in assay buffer in a final volume of 500 µl at 24°C for 3 hours to reach equilibrium conditions. Experiments were terminated by rapid vacuum filtration using a Tomtec Harvester (Tomtec, Hamden, CT), and filter-bound radioactivity was calculated by solid scintillation. Nonspecific binding was determined in the presence of 10 µM atropine.
Dissociation-Binding Assays.
To estimate ligand affinity to the allosteric binding site, [3H]NMS dissociation-binding assays were conducted. A total of 30 µg/ml membranes was preincubated with 2 nM [3H]NMS in assay buffer [Tris or Tris NaCl buffer (see above)] supplemented with 0.5% BSA and 1 µM GDP for 45 minutes at 24°C. Net dissociation of [3H]NMS was initiated by the addition of 10 µM atropine, with or without the indicated concentrations of test compound. After the appropriate time interval (12 and 9 minutes in Tris, and 6 and 3 minutes in Tris NaCl buffer), the dissociation was terminated by rapid vacuum filtration with a Tomtec Harvester (Tomtec) and filter-bound radioactivity was calculated by solid scintillation. Nonspecific binding was defined as the radioactivity bound in the presence of 10 µM atropine.
[35S]GTPγS-Binding Assay.
A total of 40 µg/ml CHO-hM2 membranes was incubated with 0.07 nM [35S]GTPγS, 1 µM GDP, 0.5% BSA, and test compound in Tris buffer (50 mM Tris, 2 mM MgCl, 1 mM Na2EDTA, 1 mM DTT, pH 7.4) or Tris NaCl buffer (50 mM Tris, 2 mM MgCl, 1 mM Na2EDTA, 1 mM DTT, 200 mM NaCl, pH 7.4) for 60 minutes at 24°C. Experiments were terminated by rapid filtration with a Tomtec Harvester (Tomtec), and filter-bound radioactivity was measured by solid scintillation counting.
Data Analysis.
Equilibrium-binding data were analyzed by a four-parameter logistic equation (Barlow and Blake, 1989) yielding IC50 values that were subsequently converted into apparent equilibrium dissociation constants (KA) using the Cheng-Prusoff correction (Cheng and Prusoff, 1973). Data from two-point kinetic dissociation experiments were analyzed using a one-phase exponential decay, as described elsewhere (Kostenis and Mohr, 1996, Voigtländer et al., 2003). Data obtained from [35S]GPTγS-binding assays were fitted to a four-parameter logistic equation yielding measures for agonist potency [negative logarithm of the concentration of test compound inducing a half maximum response (pEC50)] and maximum effect (Emax). All nonlinear regression analyses were performed using Prism 5.03 (GraphPad Software, San Diego, CA).
Statistical Analyses.
Shown are mean values ± S.E.M. Comparison of two single means was performed using an unpaired two-sample Student t test. In case the number of data sets was equal or superior to three, a one-way analysis of variance test with Bonferroni’s post-test was used.
Results
Distinct Binding Poses of Dualsteric Ligands May Result in Protean Agonism.
To unravel potential protean agonism at the muscarinic M2 receptor, we applied a variety of orthosteric agonists, allosteric fragments, and dualsteric ligands (i.e., iper-6-naph, iper-8-naph, iper-6-phth, isox-6-naph, isox-8-naph, isox-6-phth, OOM-6-naph, OOM-6-phth; Fig. 1A) and investigated receptor-mediated activation of inhibitory Gi proteins in [35S]GTPγS-binding experiments.
Test compounds and proposed mechanism of action of dualsteric ligands acting as protean agonists. (A) Structures of the dualsteric ligands examined in this study; their respective orthosteric moieties isox, iper, and OOM; and the allosteric fragments C6-naph, C6-phth, and C8-naph. The classic muscarinic partial agonist pilocarpine and the endogenous agonist ACh were used as reference compounds. (B) Proposed model to clarify the mode of action of dualsteric protean agonists at the M2 receptor. Dualsteric ligands may potentially bind either in a dualsteric binding pose, stimulating receptor signaling, or in a purely allosteric pose, inactivating the receptor (Bock et al., 2014). In this sense, dualsteric binding of protean agonists may lead to a positive signaling output in an inactive receptor system (hypothetical blue curve), whereas the compound may behave as an inverse agonist in a receptor system with spontaneous activity (hypothetical red curve).
The dualsteric ligands tested in the present study consist of an orthosteric part, that is, iper, isox, or OOM, and an allosteric moiety, that is, either the naph or phth group. The two pharmacophores are covalently connected via a hexa- or octamethylene linker. The orthosteric building block provides high affinity for interaction with the binding site of the endogenous messenger, whereas the allosteric moiety has affinity for the allosteric binding site, which is located on top of the orthosteric site in the extracellular part of mAChRs (Prilla et al., 2006; Bock et al., 2012; Haga et al., 2012). In consequence, these bipharmacophoric ligands are able to bind in two orientations to a receptor protein (Bock et al., 2014): either both pharmacophores bind to their designated binding sites, yielding a dualsteric binding pose, or the compound only interacts with the allosteric site of the receptor, yielding a purely allosteric binding pose (Bock et al., 2016) (Fig. 1B). As the allosteric fragments are derived from allosteric antagonists/inverse agonists, the purely allosteric binding pose stabilizes an inactive receptor conformation (Tränkle et al., 1998; Disingrini et al., 2006; Bock et al., 2014; Matera et al., 2014). In contrast, the dualsteric binding pose stabilizes active receptor conformations, because the orthosteric moieties are agonists (Dallanoce et al., 1999; Schrage et al., 2013; Bock et al., 2014).
The unique ability of dualsteric compounds to bind in two different binding poses leads to a potential scenario in which these model derivatives may act as protean agonists because the affinities of the two pharmacophores may change in different systems (Bock et al., 2014). A dualsteric compound that prefers binding in an allosteric binding topography will result in inverse agonism in a system with pronounced spontaneous receptor activity (red curve in lower panel of Fig. 1B). However, if this compound is also able to bind in a dualsteric binding pose, functional agonism may occur in a quiescent receptor system (blue curve in lower panel of Fig. 1B). Thus, the versatile behavior of dualsteric ligands may be an avenue toward the design of protean agonists.
Establishment of an Experimental System with a Stable Amount of Constitutive Activity of the Muscarinic M2 Receptor.
The investigation of protean agonism is technically challenging because it requires a stable, spontaneously active receptor system, which is difficult to achieve for several GPCRs (Kenakin, 2001). To establish a system that displays a robust and substantial amount of spontaneous activity for the M2AChR, we investigated the M2 receptor–mediated binding of [35S]GTPγS in CHO stably expressing the muscarinic acetylcholine M2 receptor (CHO-M2) membranes triggered by the endogenous agonist ACh or the inverse agonist atropine in the presence of various concentrations of NaCl in a Tris buffer. NaCl was chosen as a buffer supplement because sodium ions are known to be important to maintain the inactive receptor conformation of class A GPCRs (Liu et al., 2012; Katritch et al., 2014; Miller-Gallacher et al., 2014). In line with this, an increase of NaCl led to a reduction in basal [35S]GTPγS binding to CHO-M2 membranes as well as to a decrease in spontaneous receptor activity (estimated in presence of Atr) and ACh-induced [35S]GTPγS (Fig. 2Ai). To explore whether alternative conditions with an ionic strength equivalent to 200 mM NaCl may also be suitable to titrate spontaneous activity of the M2AChR, we analyzed the effect of the inverse agonist atropine sulfate on M2-mediated [35S]GTPγS binding in a Tris buffer with 2 mM sodium ions (low ionic strength Tris, Tris) in comparison with a Tris buffer substituted with 200 mM NaCl or equimolar concentrations of KCl, sodium bromide, or N-methyl-d-glucamine chloride (NMDGCl). NMDGCl is often used as a sodium substitute, as it was found to be osmotically equivalent (Pihlavisto et al., 1998; Barann et al., 2004; Billups et al., 2006; Vivo et al., 2006). Interestingly, any increase in ionic strength led to a complete abolishment of spontaneous receptor activity regardless of whether sodium or potassium salts were applied, or NMDGCl was used as a substitute (Fig. 2Aii). We conclude that conditions with an ionic strength equivalent to 200 mM NaCl may also serve to keep the receptor in the inactive state.
Establishment of a robust assay system with substantial amount of spontaneous activity for the muscarinic M2 receptor. (Ai) Increasing concentrations of NaCl affecting basal, atropine-inhibited, and ACh-induced [35S]GTPγS binding. (Aii) Effect of NaCl, KCl, sodium bromide, or NMDGCl (200 mM each) on the spontaneous activity of M2 receptors in [35S]GTPγS experiments. ***P < 0.001, significantly different from Tris according to one-way analysis of variance test with Bonferroni’s post-test. (B) Experimental dose–response curves of (Bi) ACh and (Bii) iper obtained in presence (Tris) or absence (Tris NaCl, Tris KCl) of constitutive activity of the M2 receptor. In the spontaneously active system, the whole curve was translated to higher values, as proposed in Supplemental Fig. 1B. (C) Orthosteric compound-induced [35S]GTPγS binding in CHO-hM2 membranes in absence (Tris NaCl, Ci) or presence (Tris, Cii) of spontaneous receptor activity. Although the maximum inducible effect (Emax) of isox and OOM was equal to that of ACh in both systems, iper showed a superior Emax than ACh in a silent system. (A–C) Shown are mean values ± S.E.M. of three to seven independent experiments, each performed in triplicate.
Next, we were interested in the nature of the M2 receptor system with high rates of spontaneous activity (low ionic strength Tris, Tris) in comparison with a quiescent M2 receptor system (Tris supplemented with 200 mM NaCl, Tris NaCl, or KCl, Tris KCl). We hypothesized that high spontaneous M2 activity may either do as follows: 1) increase the basal level of [35S]GTPγS binding without altering the maximum inducible effect of the system (Supplemental Fig. 1A) or 2) lead to an increase in basal [35S]GTPγS binding and also a higher potential Emax of the system (Supplemental Fig. 1B). To approach these hypotheses experimentally, we measured [35S]GTPγS binding in CHO-M2 membranes induced by the endogenous agonist ACh (Fig. 2Bi) and the muscarinic superagonist iper (Fig. 2Bii) in one system where the receptor was spontaneously active (red, Tris) and two systems where no spontaneous activity was apparent (light blue, Tris KCl, and blue, Tris NaCl). In line with our hypothesis number two (Supplemental Fig. 1B), both basal and ligand-induced receptor activity was enhanced in the spontaneously active system. We next investigated [35S]GTPγS binding to CHO-M2 membranes induced by five different orthosteric muscarinic agonists in a Tris NaCl buffer (Fig. 2Ci, quiescent receptor) and a low ionic strength Tris buffer (Fig. 2Cii, spontaneously active receptor). Agonists may prefer binding to spontaneously active receptors as this enhances the probability of ternary complex formation of agonist, receptor, and G protein (De Lean et al., 1980; Strange, 2008). Accordingly, all muscarinic agonists displayed increased potency for M2 receptor activation in a low ionic strength Tris buffer (Fig. 2C, i and ii; Supplemental Table 1), accompanied with higher apparent agonist-binding affinities (Supplemental Fig. 2; Supplemental Table 1).
Interestingly, iper shows a higher Emax value than ACh, isox, and OOM. This might indicate a lower stimulus–response coupling in the quiescent system, which enables to directly visualize the higher efficacy of iper over the other investigated agonists. Of note, the classical muscarinic partial agonist pilocarpine is an exception to the observed trend, as the Emax value and the potency did not change with the spontaneous activity of the system.
Specific Dualsteric Compounds Display Protean Agonism at Muscarinic M2 Receptors.
The ability to reliably fine-tune spontaneous M2 receptor activity enabled us to investigate [35S]GTPγS binding in CHO-M2 membranes induced by dualsteric compounds in a quiescent receptor system (Tris NaCl) or a spontaneously active receptor system (Tris). When the muscarinic superagonist iper was incorporated as the orthosteric part into the dualsteric ligand, the three tested compounds behaved as agonists in both buffer systems (Fig. 3, A–C). Iper-6-naph (Fig. 3A) and iper-6-phth (Fig. 3B) displayed higher agonist efficacy and potency in the spontaneously active receptor system in comparison with the quiescent receptor state, whereas iper-8-naph only slightly gained potency in Tris buffer over Tris NaCl (Fig. 3C). Conversely, when isox or OOM, which possess ACh-like efficacy, was applied as orthosteric building blocks (Fig. 3, D–H), the effect induced by isox-6-naph (Fig. 3D) and OOM-6-naph (Fig. 3G) changed its direction, depending on the activity state of the M2 receptor: in an inactive M2 receptor system, both compounds behaved as partial agonists (blue), whereas in a system with spontaneous activity of the receptor both displayed inverse agonism (red), as expected for protean agonists. In contrast to this and in line with the findings for iper-6-phth, isox-6-phth (Fig. 3E) and OOM-6-phth (Fig. 3H) had increased efficacy in Tris buffer as compared with Tris NaCl. Likewise, we noted higher efficacy for the isox-derived hybrid carrying an allosteric naph residue with a linker length of 8 carbon atoms (isox-8-naph) in the spontaneously active receptor system over the quiescent system (Fig. 3F). As predicted and in agreement with previous findings (Jäger et al., 2007; Antony et al., 2009; Bock et al., 2014), the allosteric fragments 6-naph, 8-naph, and 6-phth (Fig. 1A) alone behaved as allosteric inverse agonists in presence of high spontaneous M2 receptor activity but were silent at the inactive receptor (Supplemental Fig. 3).
Isox-6-naph and OOM-6-naph display protean agonism in CHO-hM2 membranes in absence (Tris NaCl) or presence (Tris) of spontaneous receptor activity. (A–C) Dualsteric ligands that carry iper as the orthosteric building block displayed partial agonism in both the silent and the spontaneously active set-up regardless of their C6-naph (A), C6-phth (B), or C8-naph (C) allosteric moiety. Noteworthily, iper-6-naph and iper-6-phth displayed significantly higher Emax values in the spontaneously active system. (D–H) Isox-6-naph, isox-6-phth, isox-8-naph, OOM-6-naph, and OOM-6-phth were all weak partial agonists in the quiescent system (blue curve) but changed their Emax in the presence of constitutive receptor activity (red curve). Isox-8-naph displayed increased receptor activation (F), isox-6-phth (E) and OOM-6-phth (H) showed no difference in effect, whereas the two compounds with a C6-naph moiety, that is, isox-6-naph (D) and OOM-6-naph (G), switched from partial to inverse agonism and were thus classified as protean ligands. (A–G) Shown are mean values ± S.E.M. of three to seven independent experiments, each performed in triplicate.
Taken together, our data allow us to derive structure–activity relationships to induce protean agonism at muscarinic M2 receptors: the orthosteric agonist should display an ACh-like efficacy for receptor activation; the allosteric residue should be bulky—the sterically demanding 1,8-naph moiety being superior to the less voluminous phth analog—and the two pharmacophores should be connected by an alkyl chain of optimally six carbon atoms.
Dissociation-Binding Experiments Point to an Increased Allosteric Affinity to M2 Receptors in Tris Buffer of a Protean Agonist.
To assess whether protean agonism of isox-6-naph and OOM-6-naph was due to an increase in allosteric binding to the [3H]NMS-bound M2 receptor in the low ionic strength buffer, we estimated binding affinities for selected compounds in the allosteric binding pose to M2. To this end, [3H]NMS dissociation experiments were performed in a low ionic strength Tris buffer (red curve) and Tris buffer supplemented with NaCl (blue curve) to measure allosteric binding affinities in the same ionic strength than the aforementioned [35S]GTPγS-binding experiments (Fig. 4).
Isox-6-naph displayed enhanced affinity to the allosteric binding site of the [3H]NMS-bound M2 receptor in Tris. (A–E) [3H]NMS dissociation experiments performed in presence of (A) iper-6-naph, (B) iper-8-naph, (C) isox-6-naph, (D) isox-8-naph, and (E) OOM-6-naph in Tris or Tris NaCl buffer. Depicted are concentration–effect curves that show the delay of [3H]NMS dissociation rate (k-1) induced by the dualsteric test compounds in two-point kinetic experiments (Kostenis and Mohr, 1996). The inflection point of the curves represents the allosteric affinity of the tested compounds for the [3H]NMS-bound receptor. (A–E) Shown are mean values ± S.E.M. of three to five independent experiments, each performed in duplicate.
Both investigated dualsteric compounds carrying iper as the orthosteric building block, that is, iper-6-naph and iper-8-naph, displayed a gain in affinity for allosteric binding (Fig. 4, A and B; Supplemental Table 1) of ∼three- and sevenfold, respectively, to M2 receptor in Tris compared with Tris NaCl buffer. The protean agonist isox-6-naph displayed a greater increase in affinity, of ∼20-fold (Fig. 4C; Supplemental Table 1), in the low ionic strength over the NaCl supplemented buffer. In contrast to this, the affinity for the allosteric binding site of isox-8-naph appeared to be similar in both buffer systems (Fig. 4D). The allosteric affinity of the OOM-containing protean agonist OOM-6-naph was significantly increased in Tris buffer in comparison with Tris NaCl (sixfold), even though the gain in affinity was not as prominent as that reported for isox-6-naph (Fig. 4E). Taken together, these data indicate that both protean agonists may display increased allosteric binding under the assay conditions similar to the spontaneously active system compared with the conditions of the inactive system and that this may contribute to the inverse agonism that we observed for isox-6-naph and OOM-6-naph.
Of note, we also observed an increase in allosteric binding for iper-6-naph and iper-8-naph to the M2 receptor in Tris compared with Tris NaCl, although these two compounds clearly did not display any protean agonism. This might be explained by the higher intrinsic efficacy for receptor activation of the orthosteric moiety iper in comparison with isox and OOM. In consequence, the beneficial effect of the spontaneously active M2 receptor on the orthosteric superagonist iper (increase in binding affinity 10-fold in Tris versus Tris NaCl) surpassed the increase in allosteric binding (increase in allosteric binding affinity 3.5-fold and sevenfold for iper-6-naph and iper-8-naph, respectively), and this precluded protean agonism.
Discussion
Protean ligands induce opposing effects (Emax) depending on the level of spontaneous activity of a GPCR. This phenomenon was first reported in 1995 (Kenakin, 1995) and is based on the assumption that ligand-bound GPCRs may adopt not only one (fully) active and one inactive conformation, but also intermediate conformations engendering intermediate intrinsic efficacies for activation of intracellular signaling pathways. In this regard, receptor species with a lower efficacy than the spontaneously active one yield positive signaling in the presence of quiescent receptors and negative signaling when the majority of receptors is constitutively active (Kenakin, 1997, 2001; Chidiac, 2002). Experimentally, the discovery of protean ligands remains challenging for two reasons, that is, the lack of the following: 1) stable and reliable spontaneously active systems for several GPCRs and 2) strategies for the rational design of this class of ligands.
In the present study, we show that spontaneous activity of the muscarinic M2 receptor can reliably be titrated by changing the ionic strength in CHO-M2 [35S]GTPγS-binding experiments. We used this system to identify dualsteric compounds as protean agonists at M2AChR. Structure–activity relationships allowed us to define the molecular features required for protean agonism at M2 receptors among a set of dualsteric compounds and to introduce a new molecular mechanism for protean agonism.
In particular, we present two dualsteric ligands, that is, isox-6-naph and OOM-6-naph, as new protean agonists at the muscarinic M2 receptor. Interestingly, only slight alterations of the three chemical moieties were sufficient to shift the profile from protean agonism to classical agonism (Fig. 3). The molecular features required for protean agonism at the muscarinic M2 receptor are as follows: an orthosteric part endowed with an acetylcholine-like efficacy, a bulky allosteric part to impair the flexibility of the extracellular loop area, and a flexible linker chain of six carbon atoms.
Up until now, protean agonism was thought to reside in the intrinsic efficacy of a ligand (Kenakin, 1997). Our results point to an additional mechanism that may underlie this phenomenon at GPCRs in case a ligand can simultaneously occupy both the binding site for the endogenous messenger and an allosteric binding site of the receptor protein. These dualsteric compounds may bind to a receptor protein in two distinct binding modes (Bock et al., 2014, 2016), a purely allosteric and a dualsteric mode (Fig. 1B), and may thus have two distinct efficacies for receptor activation. Our findings presented in this work go in line with the idea that the two protean agonists, isox-6-naph and OOM-6-naph, prefer binding in the purely allosteric binding pose, as has already been demonstrated for isox-6-naph in a previous study (Bock et al., 2014). Both compounds display functional inverse agonism under conditions in which the M2 receptor displays a substantial amount of spontaneous activity (Fig. 3, D and G). However, at least under conditions in which the receptor is silent, a significant fraction of receptors must be bound in a dualsteric binding pose, as we observed functional agonism in the quiescent M2 system. This indicates dynamic ligand binding as the molecular mechanism of protean agonism of isox-6-naph and OOM-6-naph.
Spontaneous activity of the M2AChR could robustly be fine-tuned by changing the ionic strength of the applied buffer system, such as presence/absence of sodium ions. To test whether isox-6-naph and OOM-6-naph swap their binding pose depending on the buffer system used, we performed radioligand-binding experiments and estimated the allosteric binding affinity (EC50,diss values) of several dualsteric ligands included in this study. The allosteric affinity measure of isox-6-naph was 20 times higher in Tris than in Tris NaCl buffer, whereas the orthosteric building block isox gained only 3.5-fold in affinity under these conditions (Fig. 4C; Supplemental Fig. 2). Assuming that the affinities of the allosteric and of the orthosteric moiety determine the fraction of ligand bound in one or the other pose (Bock et al., 2014), this result supports the idea that isox-6-naph increases in allosteric binding pose in the low ionic strength buffer. Yet, OOM-6-naph displayed only a slight increase in allosteric binding affinity in Tris versus Tris NaCl buffer. This effect was comparable to the increase in orthosteric affinity of the orthosteric building block OOM. Even if these data indicate that OOM-6-naph does not display increased allosteric binding in Tris buffer, the dominant binding pose under these conditions must be the allosteric pose, as we clearly see inverse agonism at the spontaneously active M2AChR (Fig. 3G). Of note, changes in the ionic strength of the buffer may not only alter the spontaneous activity of the M2AChR (see Fig. 2B, increase of the lower curve plateau), but may also affect the microenvironment in the orthosteric and allosteric binding pocket. Therefore, ligand-binding behavior may be altered not only by the functional state of the receptor protein, but additionally may be influenced by the presence/absence of ions in the binding pockets. Interestingly, all dualsteric compounds share a common bis-ammonium alkane structure that is known to engage with the allosteric M2AChR binding site via polar cation–π interactions (Dror et al., 2013), which are susceptible to variation of ionic strength (Papaneophytou et al., 2014).
However, our study clearly indicates a major impact of ionic strength on the macro level, that is, the functional state of the receptor. A recent study by Yuan et al. (2016) has shown that a continuous channel of water molecules inside the receptor protein is an essential feature of the functionally active state of the adenosine A1 receptor, another class A GPCR that most likely shares a common activation mechanism with the M2AChR. Therefore, one might speculate that ionic strength fine-tunes receptor activity by determining the amount of water molecules available within the binding pocket. Yet, albeit tempting, ultimate proof for such a hypothesis remains to be provided by crystallographic studies.
One might argue that protean agonism of isox-6-naph and OOM-6-naph resides in the rather low efficacy of the two ligands and is not necessarily due to dynamic ligand binding. However, dualsteric ligands such as isox-6-phth and isox-8-naph did not display any protean agonism, although the efficacy of these compounds did not differ significantly from the efficacy of the two protean agonists at the quiescent M2AChR. Moreover, it has been demonstrated previously that a purely allosteric binding pose predominates for isox-6-naph (Bock et al., 2014). In sum, our data indicate that dynamic ligand binding (Bock et al., 2014) is the underlying molecular mechanism for protean agonism of isox-6-naph and OOM-6-naph at M2AChR.
Protean agonism is still a seldom described phenomenon. Studies that identified protean agonists by comparison of agonist-induced signaling in two different functional assay systems may be compromised by biased signaling (Jansson et al., 1998; Gbahou et al., 2003). Therefore, two experimental systems that differ only in the amount of spontaneous activity are essential for the identification of protean agonists. To this end, several strategies have been employed to study agonism in presence and absence of spontaneously active receptors. For instance, constitutively activating receptor mutations (Ganguli et al., 1998; Fathy et al., 1999; Pauwels et al., 2002), an increased expression of receptor/G protein (Jakubík et al., 1998), or a change in buffer composition (Newman-Tancredi et al., 2003) have been applied. In this study, we chose to alter M2 wild-type receptor activity by variation in the concentration of sodium ions, because previous studies demonstrated that sodium ions can keep class A GPCRs in an inactive conformation (Liu et al., 2012; Katritch et al., 2014). Yet, our results indicate that inactivation of the M2AChR by a high concentration of sodium ions depends on high ionic strength in the buffer rather than the nature and/or composition of a particular salt. In a low ionic strength Tris buffer, basal as well as agonist-induced [35S]GTPγS incorporation was increased in CHO-M2 membranes in comparison with a buffer system with high concentrations of sodium or potassium ions. This is in line with previous findings (Tian et al., 1994) for the adrenergic α2-AR and reflects a stronger stimulus–response coupling and ternary complex formation of agonist, receptor, and G protein of the spontaneously active M2 receptor compared with quiescent M2. Accordingly, the distinct efficacies of ACh and iper could only be detected in the inactive receptor system (Fig. 2C). Moreover, almost all dualsteric compounds that did not display protean agonism showed increased Emax values, and all orthosteric agonists gained binding affinity and potency in Tris compared with Tris NaCl buffer, as agonists prefer binding to an active rather than an inactive receptor (De Lean et al., 1980). The classic muscarinic partial agonist pilocarpine is an exception to this behavior as the compound changed neither in potency nor in efficacy depending on the buffer composition. A previous work (Tota and Schimerlik, 1990) reported that pilocarpine does not distinguish between active and inactive receptors. However, direct labeling of active M2 receptors with a radioagonist indicated higher apparent affinity of pilocarpine for this receptor population in comparison with inactive receptors, albeit the affinity difference was lower than the respective difference for full agonists (Schrage et al., 2014). Together with the finding that pilocarpine activates M2 receptors in native tissue as well as in recombinant cells (Seemann et al., 2017), this implicates that pilocarpine has higher affinity for the active M2 receptor relative to the inactive form, although the nominal difference in affinity appears to be lower than for full agonists.
On the whole, we established an experimental system that allowed us to identify two new protean agonists at M2 muscarinic acetylcholine receptors. We propose an unprecedented molecular mechanism by which protean agonism is induced at a class A GPCR, that is, the M2AChR, which may virtually be extended to generate protean agonists for other class A GPCRs harboring allosteric binding sites, such as adenosine or dopamine receptors. It will be interesting to see whether this dualsteric approach will furnish protean ligands for other receptor types.
Similar to the great potential of biased GPCR signaling, improved understanding of protean agonism may provide another level toward the targeted exploitation of the GPCR signaling machinery, which could be relevant to the knowledge-based design of innovative drug candidates.
Authorship Contributions
Participated in research design: De Min, Bock, Mohr, Schrage.
Conducted experiments: De Min.
Contributed new reagents or analytic tools: Matera, Kloeckner, Muth, De Amici, Holzgrabe, Dallanoce.
Performed data analysis: De Min, Bock, Holze, Traenkle.
Wrote or contributed to the writing of the manuscript: De Min, Bock, De Amici, Kenakin, Holzgrabe, Kostenis, Mohr, Schrage.
Footnotes
- Received October 24, 2016.
- Accepted February 2, 2017.
↵1 Current affiliation: Institute for Bioengineering of Catalonia, Parc Cientific de Barcelona, Barcelona, Spain.
↵2 Current affiliation: CNS Discovery Research, UCB Pharma, Braine-l'Alleud, Belgium.
This work was supported by Deutsche Forschungsgemeinschaft (DFG) [Grants HO1368/12-1 and MO821/2-1 to K.M. and U.H.]. A.D.M. was supported by the DFG-funded Research Training Group 1873. The University of Milan financed a 4-year postdoctoral position for C.M.
↵
This article has supplemental material available at molpharm.aspetjournals.org.
Abbreviations
- [3H]NMS
- [3H]N-methylscopolamine
- [35S]GTPγS
- guanosine 5′-O-(γ-[35S]thio)triphosphate
- ACh
- acetylcholine
- BSA
- bovine serum albumin
- CHO
- Chinese hamster ovary
- CHO-hM2
- CHO stably expressing the human muscarinic acetylcholine M2 receptor
- DTT
- dl-dithiothreitol
- Emax
- maximum effect
- GPCR
- G protein–coupled receptor
- iper
- iperoxo
- KCl
- potassium chloride
- mAChR
- muscarinic acetylcholine receptor
- M2AChR
- mAChR muscarinic acetylcholine receptor 2
- MgCl
- magnesium chloride
- naph
- naphthalimide
- NMDGCl
- N-methyl-d-glucamine chloride
- OOM
- oxo-oxotremorine M
- phth
- phthalimide
- Copyright © 2017 by The American Society for Pharmacology and Experimental Therapeutics
References
In this issue
Designed Dualsteric Ligands Act as Protean Agonists at M2
