Novel M₂‐selective, G_i‐biased agonists of muscarinic acetylcholine receptors

Abstract

Background and Purpose

More than 30% of currently marketed medications act via GPCRs. Thus, GPCRs represent one of the most important pharmacotherapeutic targets. In contrast to traditional agonists activating multiple signalling pathways, agonists activating a single signalling pathway represent a new generation of drugs with increased specificity and fewer adverse effects.

Experimental Approach

We have synthesized novel agonists of muscarinic ACh receptors and tested their binding and function (on levels of cAMP and inositol phosphates) in CHO cells expressing individual subtypes of muscarinic receptors, primary cultures of rat aortic smooth muscle cells and suspensions of digested native tissues from rats. Binding of the novel compounds to M₂ receptors was modelled in silico.

Key Results

Two of the tested new compounds (1‐(thiophen‐2‐ylmethyl)‐3,6‐dihydro‐2H‐pyridinium and 1‐methyl‐1‐(thiophen‐2‐ylmethyl)‐3,6‐dihydro‐2H‐pyridinium) only inhibited cAMP synthesis in CHO cells, primary cultures, and native tissues, with selectivity for M₂ muscarinic receptors and displaying bias towards the G_i signalling pathway at all subtypes of muscarinic receptors. Molecular modelling revealed interactions with the orthosteric binding site in a way specific for a given agonist followed by agonist‐specific changes in the conformation of the receptor.

Conclusions and Implications

The identified compounds may serve as lead structures in the search for novel non‐steroidal and non‐opioid analgesics acting via M₂ and M₄ muscarinic receptors with reduced side effects associated with activation of the phospholipase C signalling pathway.

Abbreviations

IP_X

inositol phosphates

KHB

Krebs HEPES buffer

NMS

N‐methylscopolamine

VTA

ventral tegmental area

What is already known

• Highly selective muscarinic receptor agonists are needed to avoid undesired side effects.

• Only few functionally selective muscarinic agonists are known.

What this study adds

• We have developed M₂‐selective, G_i‐biased agonists.

What is the clinical significance

• These compounds may provide lead structures in the search for novel non‐steroidal and non‐opioid analgesics.

1. INTRODUCTION

Although individual GPCRs preferentially couple to a particular class of G‐proteins, their specificity is not absolute. They usually activate multiple G‐proteins with a range of potency, efficacy, and kinetics (Hermans, 2003; Jakubík, El‐Fakahany, & Dolezal, 2006). Besides artificial systems with overexpressed GPCRs, this coupling promiscuity was also observed in native cells (Jakubík et al., 1996; Laugwitz et al., 1996). It is generally accepted that structurally diverse agonists induce different changes in conformation of GPCRs that can lead to non‐uniform modulation of signalling pathways. This preferential orientation of signalling of a given GPCR towards a subset of its signal transducers is termed signalling bias (Kenakin & Christopoulos, 2013). The property may be employed to develop drugs that selectively produce desired effects while avoiding side effects associated with activation of unwanted signalling pathways (Manglik et al., 2016). Much research has been carried out on agonists biased towards either G‐protein‐ or arrestin‐mediated pathway. Here, as a proof of concept, we demonstrate signalling bias among individual classes of G‐proteins.

Five subtypes of https://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=2 are expressed in the CNS and peripheral nervous system, as well as innervated tissues. Muscarinic receptors mediate a wide variety of physiological functions via G_q/11 G‐protein‐mediated stimulation of PLC (https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=13, https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=15, and https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=17 receptors) or via G_i/o G‐protein‐mediated inhibition of AC (https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=14 and https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=16; Eglen, 2012). Disruption of muscarinic signalling contributes to many pathophysiological conditions, including a variety of neurological and psychiatric disorders, for example, Alzheimer's disease and schizophrenia, and also to some internal diseases, for example, asthma or overactive bladder (Eglen, 2012). To target these disease states, selective modulation of individual muscarinic subtypes is necessary to avoid undesired side effects. However, the high homology of the orthosteric binding site among all five muscarinic receptor subtypes (Haga et al., 2012; Kruse et al., 2012; Thal et al., 2016) makes it very difficult to find agonists that could bind and selectively activate individual subtypes. No affinity‐based selective agonists of muscarinic receptors have been discovered so far. The development of biased agonists can be the right approach to overcome the conserved nature of the orthosteric binding site and to achieve selective modulation of individual subtypes of muscarinic receptors. However, the possible signalling bias of muscarinic receptors particularly with respect to individual G‐proteins is still poorly explored.

Residues in the binding site that mediate activation of a specific signalling pathway are termed functional hot spots (Nivedha et al., 2018). Binding of an agonist to one or a subset of functional hot spots within the binding site results in activation of a subset of signalling pathways and thus in ligand‐mediated signalling bias. An agonist relatively small in size has a better chance to bind to a smaller number of functional hot spots than a larger agonist. It has been proposed that some muscarinic agonists (e.g., analogues of arecoline) exert bias towards G_i/o signalling pathway. These agonists do not interact with Asn^6.52 (Ballesteros–Weinstein numbering, Ballesteros & Weinstein, 1995) in contrast to conventional agonists (Fish et al., 2017). The scaffold containing the NCCSC backbone where the nitrogen is the electron positive part and sulphur is the electron negative part of a ligand has been shown to exert some binding selectivity for M₂ receptors (Boulos, Jakubík, Randáková, & Avila, 2013). Taking these considerations into account, we have designed and synthesized small muscarinic agonists (Figure 1). Two of these compounds exclusively inhibited synthesis of https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2352 (through activation of the G_i/o signalling pathway) and did not stimulate accumulation of inositol phosphates (IP_X; G_q pathway), either in a heterologous expression system or in digests of tissues and primary cultures. Although G_i versus β‐arrestin‐biased agonists have been reported (e.g., at https://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=50, DeWire et al., 2013), G_i/o bias at the G‐protein level has not been reported so far (Bologna, Teoh, Bayoumi, Tang, & Kim, 2017). Molecular modelling confirmed the absence of hydrogen bonding of the novel agonists with Asn^6.52 in the sixth transmembrane helix and the existence of agonist‐specific conformations induced by the novel agonists.

Figure 1.

Comparison of binding of novel agonists and agonist iperoxo to the orthosteric binding site of the M₂ muscarinic receptor. Top, chemical structures of compound 7A (left) and iperoxo (right). Bottom, compound 7A (left) and iperoxo (right) in the orthosteric binding site of M₂ receptor in an active conformation (PDB ID: 4MQS). View from extracellular side. Secondary structure of the receptor is coloured according to residue number in red–white–blue gradient TM1 being coloured red and TM7 blue. Key residues of the orthosteric binding site Asp103 (Asp^3.32) in the TM3 and Asn404 (Asn^6.52) in TM6 are displayed. Green dashed line is hydrogen bond of iperoxo to Asn⁴⁰⁴

Muscarinic receptor agonists have strong analgesic effects. Several lines of evidence indicate that the analgesic effects of muscarinic agonists are mediated by G_i/o G‐proteins activated by M₂ and M₄ receptors (Chen & Pan, 2004; De Angelis & Maria Tata, 2016; Duttaroy et al., 2002; Wess, Eglen, & Gautam, 2007). Currently, muscarinic agonists represent one of the most promising drug development targets to find novel agents to replace addictive opioid analgesics (Tata, 2008). Therefore, agonists, solely inhibiting cAMP synthesis, discovered in our laboratory, may serve as a novel pharmacophore for development of new analgesics with reduced side effects mediated by the PLC pathway. However, their potential analgesic effects will need to be verified in vivo.

2. METHODS

2.1. Cell culture and membrane preparation

CHO cells stably transfected with the genes of individual human variants of muscarinic receptors were purchased from Missouri S&T cDNA Resource Center (Rolla, MO, USA). Fresh primary cell cultures of smooth muscle cells from rat aorta, prepared as previously described (Bacáková et al., 1997) were kindly provided by Dr. Lucie Bačáková. Primary cultures of smooth muscle cells were grown to confluency in 75‐cm² flasks in DMEM supplemented with 10% FBS and cells (2 x10⁶) were subcultured up to six times. Cell cultures and crude membranes from CHO cells were prepared as described previously (Boulos, Jakubik, Boulos, Randakova, & Momirov, 2017). Cells were grown to confluency in 75‐cm² flasks in DMEM supplemented with 10% FBS, and cells (2 x10⁶) were subcultured in 100‐mm Petri dishes. Medium was supplemented with 5‐mM butyrate for the last 24 hr of culture to increase receptor expression. When needed, cells were cultured with Pertussis toxin (PTX) at a final concentration of 25 ng·ml⁻¹ for the 24 hr. Cells were detached by mild trypsinization on Day 5 after subculture. Detached cells were washed twice in 50 ml of PBS and 3‐min centrifugation at 250× g. Washed cells were suspended in 20 ml of ice‐cold incubation medium (100 mM NaCl, 20 mM Na‐HEPES, 10 mM MgCl₂, pH = 7.4) supplemented with 10 mM EDTA and homogenized on ice by two 30‐s strokes using a Polytron homogenizer (Ultra‐Turrax; Janke & Kunkel GmbH & Co. KG, IKA‐Labortechnik, Staufen, Germany) with a 30‐s pause between strokes. Cell homogenates were centrifuged for 30 min at 30,000× g. Supernatants were discarded, and pellets suspended in fresh incubation medium, incubated on ice for 30 min, and centrifuged again. The resulting membrane pellets were kept at −80°C until assayed within 10 weeks.

2.2. Preparation of rat tissue extracts

All animal care and experimental procedures complied with current legislative and guidelines and were approved by the Czech Ministry of Agriculture (permit 16OZ23113/2017‐17214). Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny, Browne, Cuthill, Emerson, & Altman, 2010) and with the recommendations made by the British Journal of Pharmacology.

Adult 8‐week‐old male Wistar rats (supplied by the Animal Facility of Institute of Physiology CAS, Prague) were killed by cervical dislocation and tissues were dissected immediately. Suspensions of digested tissue (tissue digests) from rat brain cortex, cerebellum, striatum, ventral tegmental area (VTA), and submaxillary glands were prepared as described previously by Randáková et al., (2018). Submaxillary glands were minced by scissors. Tissues were then digested by incubation for 15 min at 37°C in Krebs HEPES buffer (KHB) with trypsin (1 ml of concentrated trypsin‐EDTA solution (Sigma T4174) in 10 ml KHB), and trituration every 5 min. The resulting digest was filtered through a 40 μm cell strainer and the filtrate centrifuged for 3 min at 250× g, resuspended in KHB, and centrifuged again to remove trypsin. The tissue suspension was used immediately in experiments. Two submaxillary glands or about 400 mg of brain tissue were used in a 96‐sample assay.

2.3. Equilibrium radioligand binding experiments

All radioligand binding experiments were optimized and carried out as described by El‐Fakahany & Jakubik (2016). Briefly, cell membranes were incubated in 96‐well plates at 30°C in the incubation medium described above. Incubation volume was 400 or 800 μl for competition and saturation experiments with https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=317 (NMS) respectively. Approximately 30 μg of membrane protein per sample were used. Binding of [³H]NMS was measured directly in saturation experiments using six concentrations (30 to 1,000 pM) of [³H]NMS for 1 hr. For calculations of equilibrium dissociation constant (K _D), concentrations of free [³H]NMS were calculated by subtraction of bound radioactivity from total radioactivity in the sample and fitting Equation (1) (Section 2.5). Binding of tested ligands was determined in competition experiments with 1 nM [³H]NMS. Membranes were incubated for 1 (M₂), 3 (M₁, M₃, and M₄), or 5 hr (M₅) at 30°C. The inhibition constant K _I was calculated according to Equation (3). Non‐specific binding was determined in the presence of 1 μM unlabelled https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=320. Incubations were terminated by filtration through Whatman GF/C glass fibre filters (Whatman) using a Brandel cell harvester (Brandel, Geithesburg, MD, USA). Filters were dried in a microwave oven, and then solid scintillator Meltilex A was melted on filters (105°C, 90 s) using a hot plate. The filters were cooled and counted in Wallac Microbeta scintillation counter.

2.4. Measurements of the production of cAMP and IP_X

Levels of second messengers cAMP and IP_X were determined in radiochemical chromatography assays as described previously (Jakubík et al., 1996). To determine levels of cAMP, cells in suspension were pre‐incubated for 1 hr with 0.4 μM [³H]https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4788, washed, and incubated for 10 min in the presence of IBMX (1 mM) and 10‐μM forskolin. Then about 200,000 cells per 0.8 ml of sample were incubated for 1 hr with https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=298 or tested compounds. Incubation was ended by addition of 0.2 ml of 2.5 M HCl to the samples. Samples were applied to alumina columns (1.5 g of alumina per column), washed with 2 ml of ammonium acetate (1 M, pH = 7.0), and eluted from columns with 4 ml of ammonium acetate and measured by liquid scintillation spectrometry. M₂ and M₄ receptors that preferentially inhibit cAMP synthesis via G_i/o G‐proteins were coupled to the G_q–PLC pathway. For this purpose, new CHO cell lines stably expressing promiscuous G₁₅ G‐protein (Offermanns & Simon, 1995) and M₂ or M₄ receptors were generated by transfection with pCMV/hygro vector and hygromycin selection.

To determine the rate of formation of IP_X, suspended cells were pre‐incubated with 0.1 μM [³H]myo‐inositol for 60 min, washed, and incubated with 10 mM LiCl for 15 min. Then about 200,000 cells per 0.8 ml of sample were incubated for 1 hr with carbachol or tested compounds for 60 min. The incubation was stopped by addition of 0.5 ml of chloroform methanol HCl mixture (2:1:0.1). After separation of water and organic phases, 0.6 ml of water phase was applied to Dowex columns (1.5 g of Dowex per column), washed by 20 ml of 60‐mM ammonium formate/5 mM sodium borate buffer, and eluted from columns with 4 ml of 1 M ammonium formate/0.1 M formic acid buffer and measured by liquid scintillation spectrometry. All pre‐incubations and incubations were carried in KHB (final concentrations in mM: NaCl 138; KCl 4; CaCl₂ 1.3; MgCl₂ 1; NaH₂PO₄ 1.2; HEPES 20; glucose 10; pH adjusted to 7.4) supplemented with 10 mM glucose at 37°C.

2.5. Data and statistical analysis

The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology (Curtis et al., 2018). Experiments were independent, using different seedings of CHO cells or explanted from different animals. Binding experiments were carried out in six experiments with samples in quadruplicate, and functional assays were carried out in five experiments with samples in triplicate. Experimenters were blind to the chemical structures of tested compounds. After subtraction of non‐specific binding (binding experiments) or background/blank values (functional experiments), data were normalized to control values determined in each experiment. IC₅₀ and EC₅₀ values and parameters derived from them (K _I and K _A) were analysed after conversion to their logarithms. All data were included in analysis, no outliers were excluded, and normality of distribution was checked. In statistical analysis, value of P < .01 was taken as significant for all data. In multiple comparison tests, ANOVA with P < .01 was followed by Tukey's post‐test (P < .01). Statistical analysis was carried out using R (http://www.r-project.org) and the equations shown below.

[³H]NMS saturation binding

$$y=\frac{{\mathrm{B}}_{\mathrm{MAX}}*x}{x+K_{\mathrm{D}}},$$ (1)

where y is specific binding at free concentration x, B_MAX is maximum binding capacity, and K _D is equilibrium dissociation constant.

Competition binding

$$y=100-\frac{100*x}{x+{\mathrm{IC}}_{50}},$$ (2)

where y is specific radioligand biding at concentration x of competitor expressed as per cent of binding in the absence of competitor and IC₅₀ is concentration causing 50% inhibition of radioligand binding.Inhibition constant K _I was calculated as

$$K_{\mathrm{I}}=\frac{{\mathrm{IC}}_{50}}{1+\frac{\left[\mathrm{D}\right]}{K_{\mathrm{D}}}},$$ (3)

where IC₅₀ is concentration causing 50% inhibition of [³H]NMS binding calculated according to Equation (2) from competition binding data, [D] is concentration of [³H]NMS used, and K _D is its equilibrium dissociation constant calculated according to Equation (1) from saturation binding data.

Functional response

$$y=1+\frac{\left(\mathrm{E}{\prime }_{\mathrm{MAX}}-1\right)*x^{\mathrm{nH}}}{{\mathrm{EC}}_{50}^{\mathrm{nH}}+x^{\mathrm{nH}}},$$ (4)

where y is functional response at concentration of tested compound x, E′_MAX is apparent maximal response to the tested compound, EC₅₀ is concentration causing half‐efficient concentration, and nH is slope factor (Hill coefficient). For bi‐phasic response curves, inhibitory and stimulatory phase were fitted separately.

Operational model of functional agonism

Operational efficacy coefficient τ (Black & Leff, 1983) was determined by fitting Equation (5) to data from functional assay.

$$y=\frac{{\mathrm{E}}_{\mathrm{MAX}}*{\tau }^{\mathrm{nH}}*x^{\mathrm{nH}}}{{\left(x+K_{\mathrm{A}}\right)}^{\mathrm{nH}}+{\tau }^{\mathrm{nH}}*x^{\mathrm{nH}}},$$ (5)

where y is functional response at concentration of tested compound x, E_MAX is maximal response of the system, K _A is equilibrium dissociation constant, and nH is slope factor. Equation (5) was fitted to data from functional experiments. Equation (5) was fitted to data by the two‐step procedure described earlier (Jakubík et al., 2019). In the first step, system E_MAX was determined using https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=298, https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=302, and https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=305 as internal standards by global fit to all data for a given receptor subtype and signalling pathway. In the second step, Equation (5) with E_MAX fixed to the value determined in the first step was fitted to individual experimental data sets.

Relative intrinsic activity

For comparison of effects of agonists at different receptors and signalling pathways, relative intrinsic activity (RA_i) was calculated according to Griffin, Figueroa, Liller, and Ehlert (2007):

$${\mathrm{RA}}_{\mathrm{i}}=\frac{{\tau }_{\text{carbachol}}*K_{\mathrm{A}a}}{{\tau }_a*K_{\mathrm{A}\text{carbachol}}},$$ (6)

where τ _a and K _Aa are half‐effective concentration and apparent maximal response to the tested compound, respectively. As Hill coefficients were equal to one, RA_i values were calculated according to Equation (7).

$${\mathrm{RA}}_{\mathrm{i}}=\frac{\mathrm{E}{\prime }_{\text{MAXcarbachol}}*{\mathrm{EC}}_{50a}}{\mathrm{E}{\prime }_{\mathrm{MAX}a}*{\mathrm{EC}}_{50\text{carbachol}}},$$ (7)

where EC_50a and E′_MAXa are half‐effective concentration and apparent maximal response to the tested compound, respectively.

Signalling bias

A putative signalling bias between pathways j1 and j2 was estimated by ΔΔlog(τ/K _A) method according to Equation (8) (Kenakin & Christopoulos, 2013) based on Equation (6) (Griffin et al., 2007):

$$\text{bias}={10}^{\text{ΔΔlog}{\left(\frac{\tau }{K_{\mathrm{A}}}\right)}_{j1-j2}},$$ (8)

where

$$\text{ΔΔlog}{\left(\frac{\tau }{K_{\mathrm{A}}}\right)}_{j1-j2}=\text{Δlog}{\left(\frac{\tau }{K_{\mathrm{A}}}\right)}_{j1}-\text{Δlog}{\left(\frac{\tau }{K_{\mathrm{A}}}\right)}_{j2}.$$ (9)

2.6. Molecular modelling

2.6.1. Preparation of receptor structure

Model of the M₂ receptor in an active state was based on the crystal structure 4MQS (Kruse et al., 2013). Model of the M₂ receptor in an inactive state was based on the crystal structure 3UON (Haga et al., 2012). Non‐receptor molecules except ligand were deleted, and the resulting receptor protein was processed in Maestro using Protein Preparation Wizard according to the Sastry, Adzhigirey, Day, Annabhimoju, and Sherman (2013) guidelines.

2.6.2. Ligand docking

Docking of carbachol and compounds 6A and 7A was done using Schrödinger Software Suite (Schrödinger LLC, New York, NY) with the procedure described earlier (Jakubík, El‐Fakahany, & Doležal, 2015). The ligands were constructed in ChemAxon MarvinSketch (RRID:SCR_004111) and prepared for docking in Maestro (RRID:SCR_016748) using LigPrep (RRID:SCR_016746). The ligands were docked to the receptor using the Induced Fit Docking procedure. The M₂ receptor full agonist, https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=6937, was set to define the binding site. The Initial Glide docking step was set to “Standard Precision” without explicit constraints. In the Prime refinement step, amino acid residues within 5‐Å from the ligand were refined in a single pass. Glide re‐docking step was set to “Extra Precision.” Top 20 poses were then re‐evaluated using Prime MM‐GB/SA. The highest pose possessing hydrogen bond to Asp103 was selected for further work.

2.6.3. Simulation of molecular dynamics

To evaluate ligand binding to the receptor and to evaluate its interactions with the receptor, conventional molecular dynamics (MD) was simulated using Desmond/GPU ver. 5.3 (RRID:SCR_014575). The simulated system consisted of receptor–ligand complex in 1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine membrane set to receptor helices in water and 0.15 M NaCl. The system was first relaxed by the standard Desmond protocol for membrane proteins, and then 1.2‐ns NPT (Noose‐Hover chain thermostat at 300 K, Martyna‐Tobias‐Klein barostat at 1.01325 bar, isotropic coupling, Coulombic cut‐off at 0.9 nm) followed by 120‐ns NVE molecular dynamics without restrains was simulated. MD was run three times with various initial velocities. The quality of molecular dynamics simulation was assessed by Simulation Quality Analysis tools of Maestro and analysed by Simulation Event Analysis tool. Ligand–receptor interactions were identified using Simulation Interaction Diagram tool. Further MD trajectories were analysed in VMD (RRID:SCR_004905). Using tcl scripts (Supporting Information, Molecular Modelling section), distances and RMSDs of various parts of the receptor were measured. Namely, the distance between gravity centres of Cα of residues 94 to 124 (TM3), 385 to 411 (TM6), 204 to 212 (TM5), and 435 to 443 (TM7), all atoms of Arg121 at the edge of TM3 and Glu382 at the edge of TM6, all atoms of Tyr206 at the edge of TM5 and Tyr440 at the edge of TM7, and RMSD of Arg436, Pro437, and Tyr440 were measured. Using Dynamical Network Analysis tool box (Sethi, Eargle, Black, & Luthey‐Schulten, 2009), relay of conformational changes from the ligand to Arg121 and Glu382 via allosteric interaction networks was analysed. In this analysis, ligand and each amino acid were treated as separate nodes.

2.7. Materials

Reagents were purchased from Aldrich Chemical Company (St. Louis, MO) unless otherwise noted, and all starting liquid materials were distilled before use. NMR spectra were recorded on a Varian 300 MHz spectrometer housed at Barry University. Mass spectra were recorded on a Perkin Elmer Clarus 560 S GC/MS system. Elemental analyses were carried out by Galbraith Laboratories (Knoxville, TN), and biological assays were conducted at the Institute of Physiology of the Czech Academy of Sciences in Prague. Melting points were recorded on a MEL‐TEMP II purchased from Laboratory Devices and are uncorrected. All radiolabelled compounds ‐ (https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=317, https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4495, and https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4788) ‐ were purchased from American Radiolabelled Chemicals, Inc. (St. Louis, MO). Common chemicals were purchased from Sigma (Prague, CZ) in the highest available purity.

2.8. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20 (Alexander, Christopoulos et al., 2019; Alexander, Fabbro et al., 2019; Alexander, Mathie et al., 2019).

3. RESULTS

New compounds were tested in CHO cell lines stably expressing the human variants of individual subtypes of muscarinic receptors. Expression level of individual subtypes of muscarinic receptors was determined in [³H]NMS) saturation binding studies and is listed alongside with [³H]NMS equilibrium dissociation constants in Table S1. Expression levels of individual classes of G‐proteins was determined by [³⁵S]GTPγS scintillation proximity assay (Jakubík et al., 2006) and are summarized in Figure S1.

Affinity of new compounds (Figure S2) was assessed in competition with 1 nM [³H]NMS (Table S2). All compounds completely inhibited the binding of [³H]NMS, suggesting competitive mutually exclusive interaction. All compounds bound to all subtypes of muscarinic receptors with affinity in the micromolar range. Inhibition constants K _I ranged from 0.7 μM for compound 7D at M₂ receptors to 54 μM for compound 6A at M₄ receptors. Compounds with methylated nitrogen (7A, 7B, and 7C) had higher affinity than non‐methylated counterparts (6A, 6B, and 6C) at all receptor subtypes. Affinities of compounds with halogenated thiophene at the 5‐position (7C and 7D) ring were higher than those of non‐halogenated congeners, especially at M₂ receptors. About 20% binding to M₂ and 10% binding to M₄ of compounds 6B and 7B occurred with high affinity (about 30 nM). Reference agonists displayed high and low affinity binding (Table S3).

M₁, M₃, and M₅ receptors preferentially activate PLC and increase the level of IP_X via coupling with G_q/11 G‐proteins. M₂ and M₄ receptors preferentially inhibit production of cAMP via coupling with G_i/o G‐proteins. The ability of tested compounds to activate preferential signalling pathways of muscarinic receptors was determined in functional experiments by measuring accumulation of IP_X at M₁, M₃, and M₅ receptors or inhibition of forskolin‐stimulated production of cAMP at M₂ and M₄ receptors. These results were then compared to responses evoked by the classical non‐selective orthosteric agonists, carbachol, https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=302, and https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=305 (Table S4). Except for compound 6E (methyl group at the 4‐position of the thiophene ring), all compounds acted as partial to full agonists at all receptor subtypes, although in many cases, especially at M₁, M₃, and M₅ receptors, their effects were negligible (Table S4 and Figure S3). In particular, compounds 6A and 7A displayed at M₂ and M₄ response (E_MAX) comparable to that of carbachol. Their potency (pEC₅₀) at M₂ and M₄ receptors was higher than at other subtypes (Figure 2). Overall, all compounds were more potent and more efficacious at M₂ and M₄ receptors than at other receptors.

Figure 2.

Preferential functional responses to agonists in CHO cells that express individual subtypes of muscarinic receptors. Accumulation of inositol phosphates (IP_X; M₁, M₃, and M₅ CHO cells) or forskolin‐stimulated cAMP (M₂ and M₄ CHO cells) at non‐transfected (open symbols) or G₁₅ G‐protein transfected cells (closed symbols) was measured after stimulation by increasing concentrations of carbachol or tested compounds. Data are expressed as folds of matched controls and are means ± SD from five independent experiments performed in triplicates

Muscarinic receptors are also able to stimulate non‐preferential signalling pathways. The ability of the two most efficacious compounds 6A and 7A to activate non‐preferential responses via muscarinic receptors was determined by measuring of forskolin‐stimulated production of cAMP at M₁, M₃, and M₅ receptors or accumulation of IP_X at M₂ and M₄ receptors co‐transfected with promiscuous G₁₅ G‐protein (Figure 3). Carbachol activated non‐preferential signalling pathways at all receptor subtypes. It increased cAMP level at M₁ and M₃ receptors (G_s) and stimulated accumulation of IP_X (G₁₅) at M₂ and M₄ receptors. At M₅ receptors, carbachol inhibited forskolin‐stimulated production of cAMP at nanomolar concentrations, while at micromolar concentrations, it stimulated cAMP production. In contrast to carbachol, compounds 6A and 7A inhibited cAMP production at M₁, M₃, and M₅ receptors but did not stimulate IP_X accumulation at M₂ and M₄ receptors. Taking together all functional responses, both preferential and non‐preferential, compounds 6A and 7A inhibited forskolin‐stimulated production of cAMP at all receptor subtypes and did not stimulate the production of cAMP at any subtype. In contrast, carbachol stimulated production of IPx and cAMP at all subtypes.

Figure 3.

Non‐preferential functional responses to agonists in CHO cells that express individual subtypes of muscarinic receptors. Non‐preferential signalling was measured, namely, accumulation of forskolin‐stimulated cAMP (M₁, M₃, and M₅) or IP_X (M₂ and M₄ CHO cells) at non‐transfected (open symbols) or G₁₅ G‐protein transfected cells (closed symbols) after stimulation by increasing concentrations of carbachol or compounds 6A and 7A. Levels of IP_X and cAMP are expressed as of matched controls. Data are means ± SD from five independent experiments performed in triplicates

To calculate the operational efficacy coefficient τ of preferential (Figure 2) as well as non‐preferential (Figure 3) functional response to tested agonists, the system E_MAX was determined from functional responses to the reference agonists carbachol, oxotremorine, and pilocarpine according to the procedure described recently (Jakubík et al., 2019). As carbachol and oxotremorine did not inhibit forskolin‐stimulated production of cAMP at M₁ and M₃ receptors, system E_MAX for these systems was determined from functional responses to pilocarpine and compounds 6A and 7A. Then Equation (5) was fitted to the experimental data. Values of the operational efficacy coefficient τ of carbachol and compounds 6A and 7A (Tables S4 and S5) are plotted in Figure 4. Intrinsic activities relative to carbachol RA_i were calculated to analyse potential selectivity of 6A and 7A (Tables 1, S4, and S5). Compounds 6A and 7A had the highest RA_i at M₂ (6A, M₂ > M₃ ≈ M₄ ≈ M₅ ≈ M₁) or M₂ and M₄ receptors (7A: M₂ > M₄ > M₁ = M₃ ≈ M₅; Table 1). The reference agonist carbachol was efficacious at stimulation of IP_X (G_q) and cAMP (G_s) production at all subtypes and in lowering cAMP level at M₂, M₄, and M₅ receptors. Compounds 6A and 7A had no G_s efficacy at any receptor subtype. Their G_q efficacy was always lower than the efficacy of carbachol. In contrast, their G_i efficacy was (except M₄) always higher than efficacy of carbachol. Being G_i efficacious at M₁ and M₃ receptors, compounds 6A and 7A display an absolute G_i bias, while carbachol has no G_i efficacy at these receptors. Being solely efficacious at G_i pathway at M₂ and M₄ receptors, compounds 6A and 7A are also G_i biased at these receptors (Tables 1, S4, and S5). At M₅ receptors, compounds 6A and 7A exerted about 10‐fold and 30‐fold G_i over G_q + G₁₅ bias, respectively.

Figure 4.

Polar plot of operational efficacy coefficient τ. Operational efficacy coefficient τ was determined from functional response curves in Figures 2 and 3 by fitting Equation (5). Logarithms of τ values of inhibition of forskolin‐stimulated cAMP level (G_i), cAMP stimulation (G_s), and IP_X stimulation at non‐transfected (G_q) or G₁₅ G‐protein transfected cells (G_q + G₁₅) of reference compound carbachol and compounds 6A and 7A are plotted

Table 1.

Efficacies of compounds 6A and 7A to elicit preferential response

	6A		7A
	RAi	Δlog(τ/K A)	RAi	Δlog(τ/K A)
M1	0.0010 ± 0.0001	2.98 ± 0.26	0.0111 ± 0.0002	1.96 ± 0.16
M2	0.067 ± 0.007[Link]	1.22 ± 0.08[Link]	0.140 ± 0.008[Link]	0.93 ± 0.09[Link]
M3	0.0058 ± 0.0005	2.23 ± 0.23	0.011 ± 0.001	1.94 ± 0.20
M4	0.0046 ± 0.0004	2.15 ± 0.16	0.0209 ± 0.0011	1.58 ± 0.22
M5	0.0025 ± 0.0008	2.60 ± 0.82	0.0007 ± 0.0002	3.17 ± 0.88

Note. Stimulation of IP_X production at M₁, M₃, and M₅ receptors and inhibition of cAMP production at M₂ and M₄ receptors are expressed as intrinsic activities relative to carbachol (RA_i) and as Δlog(τ/K _A) to carbachol. Values are means ± SD from five independent experiments performed in triplicates.

*P < .01, significantly different from other subtypes; ANOVA and Tukey's post‐test.

To exclude the possibility that activation of the non‐preferential G_s pathway was obscured by activation of the preferential G_i/o pathway, accumulation of cAMP was measured in CHO cells expressing M₂ or M₄ receptors where G_i/o G‐proteins were inactivated by PTX treatment (Figure 5). PTX‐treated cells were cultured alongside with untreated cells. PTX treatment was successful as it abolished carbachol‐mediated decrease in cAMP. In fact, carbachol caused about twofold increase in forskolin‐stimulated (Figure 5) as well as basal (Figure 4) accumulation of cAMP in PTX‐treated cells. Carbachol EC₅₀ values in PTX‐treated cells were the same (M₂) or three times lower (M₄) than in non‐treated cells. Unlike carbachol, compounds 6A and 7A did not change cAMP levels even after inactivation of G_i/o G‐proteins by PTX. Thus, potential stimulatory (e.g., G_s‐mediated) response to compounds 6A and 7A was not obscured by activation of G_i G‐proteins.

Figure 5.

Accumulation of cAMP at PTX‐treated CHO cells. Accumulation of cAMP was measured in CHO cells expressing M₂ (left) or M₄ receptors (right) that had been treated with Pertussis toxin (PT_X). Cells were stimulated by increasing concentrations of carbachol, compound 6A, or compound 7A in the presence (full symbols) or absence (open symbols) of 10 μM forskolin. Level of cAMP is expressed as folds of matched controls. Data are means ± SD from five independent experiments performed in triplicates

To exclude the possibility that the observed effects of compounds 6A and 7A are the result of overexpression in the heterologous system, functional response to the compounds was measured ex vivo, in digests of normal rat tissues. Native tissues express about 10 times less muscarinic receptors than CHO cell lines (Table S6 vs. Table S1). Accumulation of IP_X was measured in brain cortex, submaxillary gland, and VTA as brain cortex and submaxillary glands are rich in M₁ and M₃ receptors, respectively, and VTA is the only region expressing significant amounts of M₅ receptors. Accumulation of cAMP was measured in M₂‐rich cerebellum and M₄‐rich striatum. Carbachol induced a marked accumulation of IP_X in the cortex, submaxillary gland, and VTA. In contrast to carbachol, compounds 6A and 7A did not increase IP_X levels in submaxillary gland and led to negligible increases in IP_X in the cortex and VTA (Figure 6, left). In the cerebellum and striatum, carbachol caused a transient decrease in forskolin‐stimulated cAMP level at sub‐micromolar concentrations followed by an increase at micromolar concentrations. In contrast to carbachol, compounds 6A and 7A caused only a decrease in cAMP level. E_MAX of inhibition of cAMP synthesis in the cerebellum and striatum by compounds 6A and 7A was comparable to inhibition by carbachol. Taken together, compounds 6A and 7A only inhibited cAMP synthesis in digests of tissues, and were sometimes as efficacious as the full agonist carbachol.

Figure 6.

Functional responses to agonists in digests of normal rat tissues. Accumulation of IP_X in brain cortex, submaxillary gland (submax gland), and ventral tegmental area (VTA) or forskolin‐stimulated cAMP in the cerebellum and striatum, after stimulation by increasing concentrations of carbachol or compounds 6A and 7A was measured. Levels of IP_X and cAMP are expressed as folds of matched controls. Data are means ± SD from five independent experiments performed in triplicates

Effects of compounds 6A and 7A on both cAMP and IP_X levels were also measured in primary cultures of rat aorta smooth muscle cells that naturally express mainly M₂ and to a lesser extent M₃ receptors (Eglen, 2012). In this primary cell culture, carbachol caused a transient decrease in cAMP level (Figure 7, left) as well as an increase in IP_X level (Figure 7, right). As in digests of normal tissues, compounds 6A and 7A only inhibited cAMP synthesis. Maximal inhibition of cAMP synthesis by compounds 6A and 7A was the same as inhibition by carbachol. Taken together, ex vivo results are in agreement with results obtained at CHO cells. Thus, the pharmacological profile of 6A and 7A was not an artefact of heterologous high‐expression systems.

Figure 7.

Functional responses to agonists in primary culture of smooth muscle cells. Accumulation of cAMP or IP_X in primary culture of smooth muscle cells was measured after stimulation by increasing concentrations of carbachol or compounds 6A and 7A. Levels of IP_X and cAMP are expressed as folds of matched controls. Data are means ± SD from five independent experiments performed in triplicates

Binding of compounds 6A and 7A was modelled in silico. Binding of compounds 6A and 7A, the classical agonist carbachol, and the super‐agonist iperoxo to the M₂ receptor in an active conformation was modelled using the crystal structure of the M₂ receptor (PDB ID: 4MQS) and analysed by simulation of molecular dynamics (MD) of receptor–ligand complex. MD analysis showed that all agonists interact with the key residue, Asp103 (Asp^3.32 according to Ballesteros–Weinstein numbering, Ballesteros & Weinstein, 1995) in TM3 (Figure 8). Compound 6A interacts with Asp103 mainly via polar interactions and water bridges. In contrast, other modelled agonists interact with Asp103 by ionic interaction that is stronger than polar interactions and water bridging. Compounds 6A and 7A may interact with key Asn404 in TM6 via water bridges, but the super‐agonist iperoxo and the full agonist carbachol interact with Asn404 mainly via polar interactions. Additionally, interaction of compounds 6A and 7A with Trp400 (Trp^6.48) was much more frequent than the interaction of iperoxo or carbachol with this residue.

Figure 8.

Interaction of agonists with the receptor during simulation of molecular dynamics. Scheme of major interactions and histograms of iperoxo (upper left), carbachol (upper right), and compounds 6A (lower left) and 7A (lower right) interactions with the receptor were calculated from MD trajectories using Ligand Interaction Diagram in Maestro. Ligand–receptor interactions are categorized into four types: hydrogen bonds (green), hydrophobic (cyan), ionic (red), and water bridges (blue). The stacked bar charts are normalized over the course of the trajectory. Values over 1.0 are possible as some protein residues may make multiple contacts of the same type with the ligand within one time frame

Using “Dynamical Network Analysis” of MD trajectory, we analysed the network of allosteric interactions from the ligand to the receptor G‐protein interface (e.g., Arg121 and Glu382; Figure S8). Optimal path of allosteric interactions from ligand to Arg121 in the edge of the TM3 helix led through Asp103–Val106–Ala109–Met112–Leu115–Ser118 in case of iperoxo; Asn108–Met112–Leu115–Ser118 in case of carbachol; and Ser107–Val111–Leu114–Ser118 in case of compounds 6A and 7A (Figure S8, blue path). Thus, new compounds relay conformational change to Arg121 via an allosteric network that differed from that of the classical agonist and that converges at Ser118. Allosteric networks of carbachol and iperoxo converge at Met112 but are initiated by different residues within the orthosteric binding site. Optimal path of allosteric interactions from ligand to Glu382 in the edge of TM6 helix led through Asn404–Thr399–Ala 395–Ala391–Arg387–Leu384 in case of iperoxo; Cys429–Thr399–Phe396–Ile392–Ile389–Thr386 in case of carbachol; and Phe195–Trp400–Ile397–Leu393–Ile389–Val385 in case of compounds 6A and 7A (Figure S8, red path). The first suboptimal path of carbachol started at Asn404. The first suboptimal path of new compounds started at Trp400. Thus, the new compounds relay conformational change to Glu382 via allosteric networks that are all the way different from those used by classical agonists. The allosteric network of carbachol is also substantially different from that of iperoxo.

Measurements of intramolecular distances over the MD trajectory were made to analyse conformational changes induced by agonists. Individual agonists did not change the size of binding site (distance between Asp103 and Asn404) significantly. In addition, the RMSD of NPxxY switch (Arg436, Pro437, and Tyr440) was similar for all agonists. Agonist induced different degree of opening of the receptor G‐protein interface (distance of TM3 to TM6 and distance of TM5 to TM7). Respective distances in the receptor complex with modelled agonists varied from the ligand‐free receptor (apo structure, Figures 9 and S5, black). The TM3 to TM6 distance was greater (about 16 Å) in case of the receptor complex with carbachol (Figures 9 and S5, red) than in the apo structure. At the other complexes, the TM3 to TM6 distance was similar to that in the apo structure. Compound 7A (cyan) induced almost no tilt of TM6 away from its inactive conformation (blue; Figure S6). Carbachol (red) induced greater tilt of TM6 than compound 7A but smaller than iperoxo (yellow). Carbachol induced slightly small tilt and shift of TM6 in M₂ receptor (yellow) than in M₁ receptor (green; Figures S6 and S7). The TM5 to TM7 distance was much smaller (about 18 Å) in the complexes of compounds 6A and 7A (Figures 9 and S5, magenta and cyan, respectively) than in the apo structure and iperoxo or carbachol complexes (about 21 Å). Monitoring distances between residues at the edge of TMs, Arg121 in TM3 to Glu382 in TM6, and Tyr206 in TM5 to Tyr440 in TM7 gave similar results with greater variation among agonists but also greater variation within individual MD trajectory. Overall opening of the receptor G‐protein interface induced by compounds 6A and 7A was smaller than the opening induced by carbachol or iperoxo. Taken together, molecular modelling data suggest agonist‐specific binding of compounds 6A and 7A followed by agonist‐specific conformation.

Figure 9.

Simulation of MD. The distance between TM5 and TM7 (abscissa) and between TM3 and TM6 (ordinate) in individual time frames of MD of the M₂ receptor in an active conformation (4MQS) with bound agonists, iperoxo or carbachol, and compounds 6A and 7A or with an empty binding site (apo)

Comparison of cryo‐EM structures 6OIJ and 6KOI of G‐protein complexes of iperoxo bound receptors in an active conformation (Maeda, Qu, Robertson, Skiniotis, & Kobilka, 2019) shows that the outward tilt of TM6 and intracellular shift of TM5 are greater in M₁–G₁₁ complex than M₂–G_oA complex (Figure S7). The distance between TM3 and TM6 is greater at M₁ (16.9 Å) than at M₂ receptor (14.6 Å). Also, the distance between TM5 and TM7 is slightly greater at M₁ (20.8 Å) than at M₂ receptor (19.9 Å). The insertion of the C‐terminus of the G_oA α‐subunit to M₂ receptor is shallower and at different angle than the insertion of C‐terminus of G₁₁ α‐subunit to M₁ receptor. Overall, greater conformational change of the receptor is needed to accommodate the G₁₁ C‐terminus than the G_oA C‐terminus.

4. DISCUSSION

The major achievement of this study is the development of novel muscarinic agonists that are selective for M₂ receptors and exhibit bias for G_i proteins, resulting in selective inhibition of production of cAMP (Figure 10).

Figure 10.

Scheme of signalling bias of novel compounds. Comparison of activation of G_i, G_q, and G_s signalling pathways by carbachol (CBC) and compounds 6A and 7A (A, right) leading to inhibition of cAMP production (G_i) or stimulation of production of IP_X (G_q) or of cAMP (G_s). Thickness of arrows denotes efficacy: thick arrow—high efficacy, thin arrow—low efficacy, dotted arrow—no activity. Carbachol activates G_q and G_s pathways via all receptor subtypes and G_i pathway via M₂ and M₄. It has no G_i efficacy at M₁ and M₃. Compounds 6A and 7A activate G_i pathway with high efficacy via all subtypes and G_q pathway with low efficacy via M₁, M₃, and M₅. They have no G_s efficacy at any receptor subtype

The phenomenon of stabilization of a particular active conformation of a receptor and consequent stimulation of some responses but not others, has been termed signalling bias (Kenakin & Christopoulos, 2013). Recent reports on muscarinic agonists have shown compounds with only modest signalling bias (Fish et al., 2017; Masuho et al., 2015). A specific signalling pathway may be activated by the interaction of an agonist with a specific amino acid within the orthosteric binding site. Such residues are termed functional hot spots (Nivedha et al., 2018). Relatively small agonists may bind only to some, not all, functional hot spots in the binding site, and thus activate only a subset of signalling pathways that result in ligand‐mediated signalling bias.

M₂ and M₄ receptors preferentially activate the G_i/o signalling pathway and inhibit cAMP synthesis; M₁, M₃, and M₅ receptors preferentially activate the G_q/11 signalling pathway and stimulate PLC, increasing the levels of https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=4222 and intracellular calcium (Eglen, 2012). Muscarinic receptors also couple to non‐preferential G‐proteins (Jakubík et al., 2006; Michal, El‐Fakahany, & Dolezal, 2007; Michal, El‐Fakahany, & Doležal, 2014). It has been proposed that some muscarinic agonists, such as analogues of https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=296, do not interact with the asparagine residue in the sixth transmembrane α‐helix (Asn^6.52 Ballesteros–Weinstein numbering, Ballesteros & Weinstein, 1995) of the orthosteric binding site and exert bias towards G_i/o signalling pathway (Fish et al., 2017). Considering this, we synthesized small muscarinic agonists. New compounds are 1‐substituted‐1‐[(5‐substituted thiophen‐2‐yl)methyl]‐3,6‐dihydro‐2H‐pyridin‐1‐ium salts (Figure S1 and Table S1). These compounds displayed a range of potency and efficacy to stimulate muscarinic receptors, mainly being partial agonists.

In measurements of functional response to agonists, the two most efficacious compounds (6A and 7A) only inhibited the forskolin‐stimulated synthesis of cAMP at M₂ and M₄ receptors (Figure 2). In contrast to the classical non‐selective agonist carbachol, compounds 6A and 7A did not activate the G_s pathway (Figures 2 and 3). To compare agonistic effects, the operational efficacies τ (Black & Leff, 1983) and agonist equilibrium dissociation constants K _A were calculated according to a recently developed procedure (Jakubík et al., 2019; Figure 4 and Tables S4 and S5). The classical definition of bias states that “A biased agonist is a ligand which stabilizes a particular active conformation of a receptor thus stimulating some responses but not others” (Lefkowitz, 2013). In line with this definition, compounds 6A and 7A, in contrast to carbachol, only inhibited production of cAMP at M₂ and M₄ receptors displaying G_i bias (Figures 2 and 3). Similarly, at M₁ and M₃ receptors, compounds 6A and 7A inhibited cAMP production, while carbachol did not. Values of τ and K _A were used to further quantify bias using the ΔΔlog(τ/K _A) method (Kenakin & Christopoulos, 2013) that is reformulation of Equation (6) for relative intrinsic activities (RA_i; Griffin et al., 2007; Tables 1, S4, and S5). Besides its “absolute” G_i bias at M₁ through M₄ receptors, compounds 6A and 7A exerted G_i bias at M₅ receptors. RA_i relative to the reference agonist carbachol Equation (7) (Griffin et al., 2007) of preferential functional response of compounds 6A and 7A were highest at M₂ receptors, suggesting their M₂ selectivity (Tables 1, S4, and S5). Good correlation between RAi values obtained according to Equations (6) and (7) suggests that τ and K _A values derived from EC₅₀ and E′_MAX according to Jakubík et al. (2019) are correct.

G_i/o G‐proteins were inactivated by PTX treatment to exclude the possibility that preferential coupling of M₂ and M₄ receptors to G_i/o G‐proteins (Figure 5) obscured signalling via G_s G‐proteins. Unlike carbachol, compounds 6A and 7A did not activate G_s G‐proteins in PTX‐treated cells. Thus, activation of G_i G‐proteins did not obscure potential activation of G_s pathway by compounds 6A and 7A. It is expected that inactivation of G_i G‐proteins by PTX removes G_i G‐proteins from the competition of various G‐proteins for receptor and EC₅₀ of carbachol stimulatory response should decrease. However, carbachol EC₅₀ remained unchanged. This may suggest that PTX only inactivates G_i G‐proteins and G_i G‐proteins remain pre‐coupled with M₂ and M₄ receptors (Jakubík, Janíčková, Randáková, El‐Fakahany, & Doležal, 2011) after PTX treatment and still compete with G_s G‐proteins.

The heterologous expression and expression level of receptors and members of the signalling pathways, including G‐proteins and effectors, may influence coupling selectivity. To exclude contribution of this system bias, the functional response to compounds 6A and 7A was also measured ex vivo in digests of several rat tissues (Figure 6) and primary culture of rat smooth muscle cells (Figure 7). Similar to CHO cells, 6A and 7A only inhibited cAMP production (activated the G_i/o signalling pathway). The cerebellum predominantly expresses M₂, and the striatum expresses M₄ and M₂ receptors (Eglen, 2012). The expression level of muscarinic receptors in these brain tissues is more than 10 times lower than in CHO cells (Table S6). Lower expression levels should lead to lower apparent E′_MAX, especially for partial agonists (Black & Leff, 1983). However, compounds 6A and 7A inhibited cAMP production to the same extent as carbachol (Figure 6), suggesting that they are full agonists at these receptors. In contrast, efficacy to stimulate IP_X production of compounds 6A and 7A in brain cortex that predominantly expresses M₁ receptors and the submaxillary gland that predominantly expresses M₃ receptors was negligible, suggesting that compounds 6A and 7A lack significant efficacy at these receptors. About 80% of muscarinic receptors expressed by smooth muscle cells are M₂, and the rest are mainly M₃ receptors (Eglen, 2012). Compounds 6A and 7A inhibited cAMP production via the M₂ receptor to the same extent as carbachol but did not stimulate IP_X production via M₃ receptors, confirming that these compounds are full agonists at M₂ and partial agonists at M₃ receptors.

Molecular dynamics of muscarinic agonists docked to the model of the M₂ receptor in an active conformation (PDB ID: 4MQS) confirmed the absence of interaction of compounds 6A and 7A with Asn404 in the sixth transmembrane α‐helix (Asn^6.52) that is essential for binding of the super‐agonist iperoxo and the full agonist carbachol (Figures 8 and S4). Allosteric interaction networks leading from the agonist to the receptor G‐protein interface of compounds 6A and 7A were different from the allosteric network of iperoxo and carbachol (Figure S8). Receptor conformations specific to compounds 6A and 7A were further confirmed by the smaller distance between TM5 and TM7 upon binding of these compounds in comparison to iperoxo or carbachol (Figure 9). Taken together, the absence of a hydrogen bond of compounds 6A and 7A with Asn404, their differential allosteric interaction network, and finding of conformations specific for compounds 6A and 7A support the theory of functional hot spots (Nivedha et al., 2018) as well as variation among allosteric interaction networks in dependence on receptor conformation (Bhattacharya & Vaidehi, 2014; Miao, Nichols, Gasper, Metger, & McCammon, 2013). A common mechanism of GPCR activation is “opening” of the receptor G‐protein interface by moving TM6 away from TM3 and TM7 away from TM5 (Deupi & Kobilka, 2007; Weis & Kobilka, 2018). The position of the TM6 was especially implicated in coupling selectivity of rhodopsin and β₂‐https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=29s (Rose et al., 2014; Semack, Sandhu, Malik, Vaidehi, & Sivaramakrishnan, 2016). It has been shown that coupling of various GPCRs to G_s G‐proteins requires a greater degree of TM6 movement than coupling to G_i G‐proteins (Glukhova et al., 2018). Cryo–electron microscopy (cryo‐EM) structures 6OIJ and 6KOI of complexes of muscarinic receptors with G‐proteins (Maeda et al., 2019) further show that greater conformational change of the receptor is needed to accommodate C‐terminus of G₁₁ G‐protein than G_oA G‐protein (Figure S7). The molecular modelling data presented here showed that agonist‐specific conformations of compounds 6A or 7A differed from agonist‐specific conformations of iperoxo or carbachol by having less effects in increasing the distance between TM5 and TM7 than that between TM3 and TM6 (Figures 9 and S5 through S7). Molecular modelling data thus suggest that coupling selectivity can be reached in various ways. In this case, data suggest that, unlike iperoxo or carbachol interacting with Asn404, compounds 6A and 7A interact with Phe195 and/or Trp400 having a shorter lever to push TM7 away from TM5. Our findings are in accordance with the recently published structures 6OIJ and 6OIK (Maeda et al., 2019) showing also different orientation of TM5 and TM7 during interaction of the G_oA α‐subunit with the M₂ receptor when compared to interaction of the G₁₁ α‐subunit with the M₁ receptor (Figure S7). Overall, our present molecular modelling data are in a good agreement with a general rule observed in cryo‐EM structures of GPCRs (Glukhova et al., 2018). G_s and G₁₁ coupling receptors show a greater degree of G‐protein interface opening than G_i/o coupling receptors. In turn, molecular modelling data are in a good agreement with experimental data. Compounds 6A and 7A that cause a small opening of G‐protein interface in silico activate only G_i/o signalling pathway in vitro. Carbachol that causes a greater opening of the G‐protein interface in silico activates all three studied signalling pathways in vitro.

The tradition of chewing betel for its stimulating effects and pain relief dates back thousands of years. These effects of betel have been ascribed to the muscarinic agonists arecoline and https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=9487 present in areca (betel) nut. Muscarinic agonists can produce strong analgesic effects, at the peripheral, spinal, and brain levels. Analgesic effects of the muscarinic agonist oxotremorine in the tail flick and hot plate tests were largely reduced in M₂ knockout mice and absent in M₂/M₄ double knockout mice (Duttaroy et al., 2002). Several lines of evidence suggest that M₂ and M₄ receptors inhibit pain transmission at the spinal level (Chen & Pan, 2004) and suggest the potential of M₂ agonists as analgesics (Wess et al., 2007). The proposed mechanism of peripheral analgesia mediated by M₂ receptors is in principle similar to the analgesia mediated by opioid receptors. Activation of M₂ or https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=319 leads to activation of G_i/o G‐proteins. The active α‐subunits released from activated G_i/o G‐proteins inhibit AC. The consequent decrease in cAMP causes attenuation in the activity of tetrodotoxin‐resistant https://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=82. The βγ dimers released from activated G_i/o G‐proteins activate https://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=74. A decrease in the activity of sodium channels and activation of potassium channels results in a decrease in excitability of nociceptive sensory neurons (De Angelis & Maria Tata, 2016). Moreover, in a model of chronic neuropathic pain, reduction of the expression of M₂ receptors in the dorsal root ganglion was accompanied by a diminished analgesic response to muscarinic agonists (Zhang, Chen, Chen, & Pan, 2018), confirming the role of M₂ receptor in spinal nociceptive transmission. In contrast to pain, secretion of exocrine and endocrine glands and smooth muscle contraction is mediated primarily by G_q G‐proteins activated by M₃ receptors (Ehlert, 2003; Stengel, Yamada, Wess, & Cohen, 2002). Sole activation of M₂ and M₄ receptors may prevent some undesired side effects of common non‐selective muscarinic agonists, including excessive sweating and salivation and, importantly, incontinence mediated by peripheral M₃ receptors. Restricting unwanted side effects of M₂ receptor activation, for example, bradycardia or smooth muscle contraction in chronic pulmonary obstructive disease and other smooth muscle related diseases, will require either topical administration (Wess et al., 2007) or advanced approaches like targeted drug delivery or tissue‐selective action (Seemann et al., 2017). If analgesic effects are confirmed in vivo, compounds 6A and 7A may serve as lead compounds in the search for novel non‐steroidal, non‐opioid analgesics.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

A.R. conducted experiments and data analysis and contributed to experimental design and manuscript, D.N. and D.U. conducted experiments and managed cell lines, P.N. and Q.S. synthesized compounds, V.D. and E.E.E. contributed to experimental design and manuscript, J.B. managed design and synthesis of compounds and contributed to manuscript, J.J. conducted molecular modelling, managed experimental design, experiments, and data analysis, and wrote manuscript. All co‐authors approved final version of manuscript.

DECLARATION OF TRANSPARENCY AND SCIENTIFIC RIGOUR

This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for https://bpspubs.onlinelibrary.wiley.com/doi/abs/10.1111/bph.14207, and https://bpspubs.onlinelibrary.wiley.com/doi/full/10.1111/bph.14206, and as recommended by funding agencies, publishers, and other organizations engaged with supporting research.

Supporting information

Data S1 Supporting information

Randáková A, Nelic D, Ungerová D, et al. Novel M₂‐selective, G_i‐biased agonists of muscarinic acetylcholine receptors. Br J Pharmacol. 2020;177:2073–2089. 10.1111/bph.14970