Authors: J Jakubík, H Janíčková, EE El-Fakahany, V Doležal
Categories: Research Papers, GDP binding, agonist efficacy, muscarinic acetylcholine receptors
Source: British Journal of Pharmacology
Conventional determination of agonist efficacy at G-protein coupled receptors is measured by stimulation of guanosine-5′-γ−thiotriphosphate (GTPγS) binding. We analysed the role of guanosine diphosphate (GDP) in the process of activation of the M2 muscarinic acetylcholine receptor and provide evidence that negative cooperativity between agonist and GDP binding is an alternative measure of agonist efficacy.
Filtration and scintillation proximity assays measured equilibrium binding as well as binding kinetics of [^35^S]GTPγS and [^3^H]GDP to a mixture of G-proteins as well as individual classes of G-proteins upon binding of structurally different agonists to the M2 muscarinic acetylcholine receptor.
Agonists displayed biphasic competition curves with the antagonist [^3^H]-N-methylscopolamine. GTPγS (1 µM) changed the competition curves to monophasic with low affinity and 50 µM GDP produced a similar effect. Depletion of membrane-bound GDP increased the proportion of agonist high-affinity sites. Carbachol accelerated the dissociation of [^3^H]GDP from membranes. The inverse agonist N-methylscopolamine slowed GDP dissociation and GTPγS binding without changing affinity for GDP. Carbachol affected both GDP association with and dissociation from Gi/o G-proteins but only its dissociation from Gs/olf G-proteins.
These findings suggest the existence of a low-affinity agonist-receptor conformation complexed with GDP-liganded G-protein. Also the negative cooperativity between GDP and agonist binding at the receptor/G-protein complex determines agonist efficacy. GDP binding reveals differences in action of agonists versus inverse agonists as well as differences in activation of Gi/o versus Gs/olf G-proteins that are not identified by conventional GTPγS binding.
Keywords: muscarinic acetylcholine receptors, agonist efficacy, GDP binding
Almost 900 genes of the human genome encode several thousands of G-protein coupled receptors (GPCRs). GPCRs thus represent the largest family of receptors. The heterotrimeric guanine nucleotide-binding proteins (G-proteins) function to transduce signals from these receptors to effector systems including enzymes, such as adenylyl cyclase and phospholipase C and ion channels. Binding of an agonist to a GPCR induces conformational changes in the receptor protein that enable the receptor to promote guanosine diphosphate (GDP) release from the α-subunit of interacting heterotrimeric G-proteins (Gα) (Wess, 1997) and formation of a high-affinity complex with guanine nucleotide-free Gα (Kent et al., 1980). The Gα subunit dissociates from the agonist-receptor-Gα complex upon binding of GTP and releases free Gα with bound GTP and βγ dimer, both of which are involved in regulation of the activity of various effector systems.
The biological activity of an agonist is a product of both affinity and efficacy. While affinity of an agonist for a receptor is strictly given by free binding energy, agonist efficacy in transducing a signal across the cell membrane depends on time-ordered complex conformational changes involving interactions among agonist, receptor, G-protein and guanine nucleotides. These interactions and the resulting conformational changes are less well characterized. In their pioneering work, De Lean et al. (1980) reported that GDP did not affect the efficacy of β-adrenoceptor agonists at Gs G-protein-coupled receptors. However, it has been repeatedly demonstrated that GDP affects binding of agonists at Gi G-protein coupled GPCRs (Florio and Sternweis, 1989; Tota and Schimerlik, 1990), muscarinic agonists decrease GDP binding (Haga et al., 1986; Shiozaki and Haga, 1992) and accelerate its dissociation (Ferguson et al., 1986). Although the structural basis for many steps in the G-protein nucleotide cycle have been clarified over the past decade, the precise mechanism for receptor-mediated G-protein activation (GDP-GTP exchange) remains incompletely defined largely because of difficulties in obtaining crystals of receptor G-protein complexes for X-ray diffraction analysis (Johnston and Siderovski, 2007; Oldham and Hamm, 2008).
The aim of our study was to investigate in detail the mechanisms that determine efficacy of agonists at individual classes of G-proteins coupled to M2 muscarinic acetylcholine receptors in natural membrane environments. We performed detailed analyses of allosteric interactions between guanine nucleotides and four structurally distinct agonists exhibiting different potencies and efficacies at the M2 receptor expressed in Chinese hamster ovary (CHO) cells. We showed that the efficacy of these agonists in stimulation of GTP binding correlates with the magnitude of negative cooperativity with GDP binding to the receptor G-protein complex. These data suggest that the decrease in GDP affinity due to acceleration of its dissociation plays a key role in determining agonist efficacy at the muscarinic M2 receptor. We suggest that measurements of GDP binding provide additional information on receptor activation to that obtained from GTP binding assays. Most importantly, it reveals differences in the action of agonists and inverse agonists that are not observable in GTP binding studies.
Chinese hamster ovary cells stably transfected with the human M2 muscarinic receptor gene (CHO-M2 cells) were kindly donated by Professor T.I.Bonner. Cell cultures and crude membranes were prepared as described previously (Jakubík et al., 2006). Briefly, cells were grown to confluency in 75 cm^2^ flasks in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 2 × 10^6^ cells were subcultured to 100 mm Petri dishes. Medium was supplemented with 5 mM butyrate for the last 24 h of cultivation to increase receptor expression. Cells were detached by mild trypsinization on day 5 after subculture. Detached cells were washed twice in 50 mL of phosphate-buffered saline 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 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, pellets resuspended in fresh incubation medium and centrifuged again. Resulting membrane pellets were kept at −20°C until assayed within a maximum of 10 weeks.
Membrane-bound GDP was removed by mild denaturation (Ferguson et al., 1986). Membranes were incubated for 3 h in 1 M ammonium sulphate at 4°C, centrifuged and resuspended in incubation medium containing 20% glycerol for 1 h to allow renaturation. Then they were again centrifuged, resuspended in incubation medium, and used for experiments.
All radioligand binding experiments were optimized and carried out as described earlier (Jakubík et al., 2006). Briefly, membranes were incubated in 96-well plates at 30°C in the incubation medium described above that was supplemented with freshly prepared dithiotreitol at a final concentration of 1 mM. Incubation volume was 200 µL or 800 µL for [^3^H]N-methylscopolamine (NMS) saturation experiments. Approximately 30 and 10 µg of membrane proteins per sample were used for [^3^H]NMS and [^35^S]GTPγS binding respectively. NMS binding was measured directly in saturation experiments using six concentrations (30 pM to 1000 pM) of [^3^H]NMS for 1 h. Depletion of radioligand was smaller than 20% for the lowest concentration. For calculations, radioligand concentrations were corrected for depletion. Agonist binding was determined in competition experiments with 1 nM [^3^H]NMS. Membranes were first pre-incubated 60 min with agonists and guanine nucleotides, if applicable, and then incubated with [^3^H]NMS for additional 120 min. Non-specific binding was determined in the presence of 10 µM NMS. Equilibrium [3H]GDP binding was measured after 5 h incubation. Non-specific binding was determined in the presence of 50 µM GDP. Agonist stimulated [^35^S]GTPγS binding was measured in a final volume of 200 µL of incubation medium with 500 pM of [^35^S]GTPγS and 50 µM GDP for 20 min at 30°C after 60 min pre-incubation with GDP and agonist. Non-specific binding was determined in the presence of 1 µM unlabelled GTPγS. Incubations were terminated by filtration through Whatman GF/F glass fibre filters (Whatman, Maidstone, UK) using a Tomtech Mach III cell harvester (Perkin Elmer, Norwalk, CT, USA). Filters were dried in vacuum for 1 h at 80°C 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.
Kinetics of [^35^S]GTPγS binding at GDP-less membranes was measured in a final volume of 200 µL at 30°C. Association of 1 nM [^35^S]GTPγS with GDP-less membranes was measured after 20 min pre-incubation with buffer or carbachol ± 50 µM GDP. Dissociation of [^35^S]GTPγS was initiated by 1 µM GTPγS alone or in mixture with carbachol after 90 min pre-incubation of GDP-less membranes with 1 nM [^35^S]GTPγS ± 50 µM GDP. Kinetics of [^3^H]GDP binding at Gi/o and Gs/olf G-proteins was measured using scintillation proximity assay (SPA) (DeLapp et al., 1999) essentially as described earlier (Jakubík et al., 2006). Association of 500 nM [^3^H]GDP was measured after 20 min pre-incubation of GDP-less membranes with buffer or carbachol. Dissociation of [^3^H]GDP was started by addition of 50 µM GDP alone or in mixture with carbachol after prelabelling GDP-less membranes with 500 nM [^3^H]GDP for 180 min at 30°C. Dissociation was stopped by cooling and solubilization of samples by adding Nonidet P-40 to final concentration of 1% for 15 min. Primary polyclonal rabbit IgG antibody against α subunit of Gi/o or Gs/olf G-proteins in final dilution 1000 was then added and samples were incubated on ice for 60 min. Afterwards, 50 µL aliquots of anti-rabbit IgG coated scintillation beads were added (Amersham Bioscience, Buckinghamshire, UK; 500 mg of beads was resuspended in 40 mL of incubation buffer) and incubation continued for another 3 h. Trapped α subunits were pelleted at 4°C and 1500×g for 15 min and counted using SPA protocol in Wallac Microbeta scintillation counter.
In general binding data were analysed as described previously (Jakubík et al., 2006). Data were preprocessed by Open Office 2.3 (http://www.openoffice.org) and subsequently analysed by Grace 5.1.18 (http://plasma-gate.weizman.ac.il/Grace) and statistic package R (http://www.r-project.org) on Mandriva distribution of Linux.
The following equations were fitted to
Saturation of radioligand binding
y, binding of radioligand at free concentration of radioligand x; BMAX, maximum binding capacity; KD, equilibrium dissociation constant.
Concentration–response
y, radioactivity in the presence of agonist at concentration x normalized to radioactivity in the absence of agonist; EMAX, maximal increase by agonist; EC50, concentration of agonist producing 50% of maximal effect; nH, Hill coefficient.
Interference with [^3^H]NMS or [^3^H]GDP binding
y, binding of radioligand at a concentration of displacer x normalized to binding in the absence of displacer; IC50, concentration causing 50% decrease in binding; nH, Hill coefficiet; flow, percentage of low affinity sites; IC50high, concentration causing 50% decrease in binding to high affinity sites; IC50low, concentration causing 50% decrease in binding to low affinity sites. Both equations were fitted to data and the one giving better fit determined by F-test was used. Equilibrium dissociation constant of displacer (KI) was calculated according to Cheng and Prusoff (Cheng and Prusoff, 1973).
Rate of association
y, binding of radioligand at time x; kobs, kobs1, kobs2, observed rates of association; Beq, Beq1, Beq2, equilibrium binding.
Rate of radioligand dissociation
y, binding of radioligand at time x normalized to time 0; koff, koff1, koff2, rate constants; f2, percentage of sites with rate constant koff2.
Allosteric interaction between a radioligand and an allosteric modulator was analysed according to the ternary complex model (Ehlert, 1988).
y, binding of radioligand in the presence of ligand A at concentration x normalized to the absence of ligand A; [D] concentration of radioligand; KD, equilibrium dissociation constant of radioligand; KA; equilibrium dissociation constant of ligand A; α, factor of cooperativity between radioligand and ligand A.
Allosteric interaction between GDP and agonist binding was analysed according to the ternary complex model with agonists competing with radioligand (Jakubík et al., 1997).
y, binding of radioligand ([^3^H]NMS) in the presence of GDP at concentration x normalized to the absence of GDP; [D] concentration of radioligand; KD, equilibrium dissociation constant of radioligand; [A], concentration of agonist; KI, equilibrium dissociation constant of high affinity agonist binding form Eqn 3; KA, equilibrium dissociation constant of allosteric ligand (GDP); α, factor of cooperativity between radioligand and allosteric ligand from Eqn 7 (always 1); β, factor of cooperativity between allosteric ligand and agonist.
The radioligands [^3^H]NMS ([^3^H]GDP), [^35^S]GTPγS and anti-rabbit IgG-coated scintillation proximity beads were from Amersham (UK). Rabbit polyclonal antibodies against C-terminus of G-protein (Gi/o, C-10, and Gs/olf, C-18) were from Santa Cruz Biotechnology (Santa Cruz, CA, USA) Carbachol, dithiotreitol, EDTA, GDP, GTPγS, NMS chloride and pilocarpine were from Sigma (St. Louis, MO, USA). Oxotremorine was from RBI (Natick, MA, USA) and Nonidet P-40 was from USB Corporation (Cleveland, OH, USA). Furmethide was kindly donated by Dr Shelkovnikov (University of St. Petersburg, Russia). Nomenclature of receptors and G-proteins follows Alexander et al. (2009).
Experiments were performed on membranes of CHO-M2 cells stably expressing 1.4 ± 0.2 pmol of binding sites for [^3^H]NMS chloride per mg of membrane protein. The equilibrium dissociation constant (KD) of [^3^H]NMS was 512 ± 32 pM (mean ± SEM, n= 4, measurements on cells from independent seedings). Total binding of [^35^S]GTPγS to crude membranes was 123 ± 18 pmol per mg of protein out of which 74 ± 11 pmol was to Gi/o, 21 ± 3 pmol to Gq/11 and 16 ± 3 pmol to Gs/olf G-proteins respectively (means ± SEM, n= 3).
Carbachol, furmethide, oxotremorine and pilocarpine concentration dependently stimulated binding of [^35^S]GTPγS (Figure 1, Table 1). Carbachol and furmethide induced similar maximal increase of [^35^S]GTPγS binding (threefold and 3.1-fold increase respectively) with half-effective concentrations (EC50) of 12.3 and 7.0 µM respectively. Oxotremorine and pilocarpine were more potent (EC50= 1.0 and 1.2 µM respectively) but less efficacious (Emax= 2.8 and 1.6-fold increase respectively). The rank order of efficacy furmethide = carbachol > oxotremorine > pilocarpine, with a ranking of potency oxotremorine = pilocarpine > furmethide = carbachol. Carbachol had no effect on [^35^S]GTPγS binding at wild-type (non-transfected) CHO cells.
Figure 1 Stimulation of [^35^S]guanosine-5′-γ−thiotriphosphate (GTPγS) binding by agonists. [^35^S]GTPγS binding to membranes stimulated by increasing concentrations (abscissa, log M) of agonists carbachol, furmethide, oxotremorine, pilocarpine and antagonist N-methylscopolamine (NMS) is expressed as fold over basal (ordinate). Data are mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equation 2 and results of fits are shown in the Table 1.
Affinity of agonist binding was assessed indirectly in competition experiments with 1 nM of the muscarinic radioligand [^3^H]NMS (Figure 2). Competition curves were biphasic and displayed a similar proportion (50 to 66%) of low-affinity binding sites for all agonists but different affinities for both high- and low-affinity binding sites (Table 2). High-affinity binding ranged from 12 nM for oxotremorine to 120 nM for carbachol and low-affinity binding from 580 nM for oxotremorine to 9 µM for carbachol. Competition curves between [^3^H]NMS and agonists in the presence of 1 µM GTPγS expectedly became monophasic for all agonists (Figure 2) with calculated equilibrium inhibition constants (KI) corresponding to the low-affinity KI in the absence of GTPγS. Similarly, 50 µM GDP present during competition measurements (Figure 2) also converted curves to monophasic ones with KI corresponding to that in the presence of GTPγS and the low-affinity KI in the absence of added nucleotides (Figure 2, Table 2).
Figure 2 Effects of guanine nucleotides on competition between agonists and [^3^H]N-methylscopolamine (NMS) binding. Binding of 1 nM [^3^H]NMS to membranes in the presence of increasing concentrations (abscissa, log M) of the agonists carbachol (upper left), furmethide (upper right), oxotremorine (lower left) and pilocarpine (lower right) is expressed as per cent of control binding in the absence of agonist. Data are shown for binding to crude membranes (control); binding to crude membranes in the presence of 50 µM guanosine diphosphate (GDP); binding to crude membranes in the presence of 1 µM guanosine-5′-γ−thiotriphosphate (GTPγS); and binding to GDP-less membranes. Data are mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equations 3 and 4. Results of fits are shown in the Table 2.
In order to explore the effects of added GDP on agonist binding, we reduced membrane-bound native GDP by inducing its dissociation under slightly denaturing conditions, washing and renaturing as described in Methods. Competition curves remained biphasic (Figure 2) but the proportion of low-affinity sites decreased fivefold to sevenfold compared with membranes before treatment (Table 2). Low-affinity KI of agonists corresponded to the low-affinity KI under control conditions. High-affinity KI values were significantly lower for carbachol and oxotremorine (twofold and fourfold respectively) and not changed for oxotremorine and pilocarpine.
In comparison with crude membranes, depletion of GDP did not change the affinity for [^3^H]NMS (498 ± 29 pM) but increased the number of binding sites per mg of protein to 24 ± 3 pmol (mean ± SEM, n= 4). On the other hand, total binding of [^35^S]GTPγS per mg of protein fell to 28 ± 2 pmol out of which 18 ± 1 pmol was to Gi/o, 3.5 ± 0.5 pmol to Gq/11 and 4.6 ± 0.6 pmol to Gs/olf G-proteins respectively (means ± SEM, n= 3).
Measurements of 1 nM [^35^S]GTPγS binding kinetics were carried out on GDP-depleted membranes without (Figure 3, left) or with added 50 µM GDP (Figure 3, right). As shown in Table 3, addition of GDP slowed down the rate of association 3.2-fold, decreased equilibrium binding 4.7-fold, but did not change the rate of dissociation (Figure 3, Table 3). A saturating concentration of carbachol (100 µM) did not change kinetics of [^35^S]GTPγS binding in the absence of GDP (Figure 3, left, open circles). In the presence of GDP, carbachol had no effect on the rate of dissociation of [^35^S]GTPγS but accelerated the rate of association 2.6-fold and increased equilibrium binding fourfold (to 92 and 90% of that in the absence of GDP respectively). In the presence of GDP, the inverse agonist NMS slowed the association of [^35^S]GTPγS by 25% and, similarly to carbachol, did not change dissociation kinetics (Table 3).
Figure 3 Kinetics of [^35^S]guanosine-5′-γ−thiotriphosphate (GTPγS) binding to membranes. Time course of association of 1 nM [^35^S]GTPγS with (top row) and dissociation from (bottom row) GDP-less membranes. Binding was carried out either in the absence (left graphs) or presence (right graphs) of 50 µM GDP in the absence or presence of 100 µM carbachol or 100 nM N-methylscopolamine (NMS). Data are presented as mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equations 5 (association) or 6 (dissociation). Results of fits are shown in the Table 3.
Measurements of 500 nM [^3^H]GDP binding kinetics (Figure 4) were carried out on GDP-less membranes in the absence or in the presence of 10 µM or 100 µM carbachol or 0.1 µM NMS. Association of [^3^H]GDP was biphasic with an observed association rate (kobs slow) of 0.010 min^−1^ for 44% of sites and kobs fast of 0.063 min^−1^ for the rest. Ten µM carbachol decelerated association of the slower fraction sevenfold while it slowed down that at the faster fraction by only twofold. Carbachol (100 µM) brought further slowing down of both the slower and faster fractions to 0.00024 min^−1^ and 0.013 min^−1^ respectively. Dissociation of [^3^H]GDP was also biphasic with dissociation rate constants koff slow 0.85 min^−1^ for 40% of sites and koff fast 0.073 min^−1^ for the rest. Carbachol accelerated [^3^H]GDP dissociation from the faster fraction more than 100-fold at both concentrations while the rate at the slower fraction was reduced threefold and fourfold at 10 and 100 µM carbachol, respectively.
Figure 4 Kinetics of [^3^H]GDP binding to membranes. Time course of association 500 nM [^3^H]GDP with (left) and dissociation from (right) GDP-less membranes in the absence or presence of 10 µM or 100 µM carbachol or 100 nM N-methylscopolamine (NMS). Data are expressed as mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equations 5 (association) or 6 (dissociation).
Carbachol-induced changes in GDP kinetics confirmed allosteric interactions between GDP and carbachol. In order to quantify allosteric interactions between GDP and agonists, their affinities to free receptor-G-protein complex had to be known. Affinity of GDP was determined in equilibrium binding of [^3^H]GDP to GDP-less membranes in homologous competition (Figure 5 left) and saturation (Figure 5 right) experiments. Homologous competition curves of 1 µM and 5 µM [^3^H]GDP were monophasic with Hill coefficient equal to 1 and IC50 values 4.48 (95% interval of confidence 3.93–5.02) and 8.55 (95% interval of confidence 7.62–9.59) µM, respectively, giving KD for [^3^H]GDP of 3.49 µM. In accordance with competition experiments, saturation binding of GDP-less membranes with 0.3 to 10 µM [^3^H]GDP displayed KD of 3.47 ± 0.03 µM and BMAX of 28 ± 3 fmol of binding sites per µg of protein.
Figure 5 Equilibrium binding of [^3^H]GDP to membranes. Left: Homologous competition of GDP (abscissa, log M concentration of GDP) with 1 µM and 5 µM [^3^H]GDP binding (ordinate, percent of control binding). Right: [^3^H]GDP saturation binding (abscissa, concentration in µM; ordinate, [^3^H]GDP binding in fmol·µg^−1^ protein). Data are presented as mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equations 1 and 3 as appropriate.
In the first set of experiments to quantify the magnitude of allosteric interactions between GDP and agonists, the binding of [^3^H]GDP at fixed 10 µM concentration and increasing concentrations of tested agonists was measured in competition-like experiments (Figure 6). While the inverse agonist NMS did not affect [^3^H]GDP binding, all the tested agonists decreased it. Fitting Eqn 7 to data using all four agonists gave the same equilibrium dissociation constant for [^3^H]GDP (pKA 5.51 ± 0.05; mean ± SD; n= 12). Factors of cooperativity α between binding of [^3^H]GDP and carbachol, furmethide, oxotremorine and pilocarpine were (expressed as pα) −2.3 ± 0.2, −2.4 ± 0.3, −1.8 ± 0.1 and −1.4 ± 0.1 (mean ± SEM., n= 3) respectively.
Figure 6 Direct measurement of allosteric interactions between agonists and [^3^H]GDP at membranes. The magnitude of allosteric interactions between agonists (carbachol, furmethide, oxotremorine, pilocarpine) or antagonist [N-methylscopolamine (NMS)] and GDP was measured directly as changes in equilibrium binding of 10 µM [^3^H]GDP to GDP-less membranes in the presence of increasing ligand concentration (abscissa, log M). Data are expressed as mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equation 7.
In the second set of experiments determining the magnitude of allosteric interactions between GDP and agonists, we employed [^3^H]NMS as a tracer because of the high cost of [^3^H]GDP and difficulties in quantifying negative cooperativity between GDP and full agonist (carbachol, furmethide) binding. The magnitude of negative cooperativity between agonists and GDP in these experiments was derived from a decrease in displacement of [^3^H]NMS binding by a fixed concentration of tested agonist by increasing concentrations of GDP in GDP-less membranes. In the absence of agonist, GDP had no effect on [^3^H]NMS binding (Figure 7). Eqn 7 could not be fitted to the data and the factor of cooperativity α between [^3^H]NMS and GDP is thus equal to 1. Agonists competed with [^3^H]NMS and diminished its binding (Figure 7). GDP allosterically reduced the affinity for agonists that was manifested as an increase in [^3^H]NMS binding. Factors of cooperativity β between GDP and agonists were calculated by fitting Eqn 8 to the data shown in Figure 7. GDP diminished the affinity of the full agonists furmethide and carbachol 250-fold and 200-fold, respectively, while the affinity of the partial agonists oxotremorine and pilocarpine was reduced only 60-fold and 25-fold respectively. Estimated affinity for GDP was 3.2 µM (pKA=−5.49 ± 0.03; mean ± SEM.; n= 12) for all fits.
Figure 7 Indirect measurement of allosteric interactions between agonists and GDP at membranes. The magnitude of allosteric interactions between agonists (carbachol, upper left; furmethide, upper right; oxotremorine, lower left; pilocarpine, lower right) and GDP was measured indirectly as changes in equilibrium binding of 1 nM [^3^H]N-methylscopolamine (NMS) to GDP-less membranes in the presence of a fixed concentration of agonist and increasing concentrations of GDP (abscissa, log M). Binding of 1 nM [^3^H]NMS in the absence of agonis is also shown (control). Carbachol and furmethide were used at 3 and 10 µM, oxotremorine at 0.1 and 0.3 µM, and pilocarpine at 0.3 and 1 µM. Data are presented as mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equations 7 ([^3^H]NMS alone) or 8 ([^3^H]NMS in the presence of agonist).
Binding of [^35^S]GTPγS to individual subclasses of G-proteins was measured in SPAs (Figure 8). As expected, all tested agonists stimulated binding of [^35^S]GTPγS to Gi/o G-proteins with higher potency than to Gs/olf and Gq/11 G-proteins (Table 4). The rank order of potencies was oxotremorine = pilocarpine > furmethide > carbachol at all tested G-protein subclasses (except no stimulation of [^35^S]GTPγS binding by oxotremorine at Gq/11 was detected). Agonists were also more efficacious in stimulating [^35^S]GTPγS binding at Gi/o G-proteins than at the other two G-protein classes. The rank order of agonist efficacies to stimulate [^35^S]GTPγS binding varied among G-protein classes and was as furmethide = carbachol = oxotremorine > pilocarpine at Gi/o, carbachol = furmethide > oxotremorine > pilocarpine at Gs/olf, and carbachol > furmethide > pilocarpine > oxotremorine at Gq/11.
Figure 8 Stimulation of [^35^S]guanosine-5′-γ−thiotriphosphate (GTPγS) binding to G
i/o, Gs/olfand Gq/11G-proteins by agonists. [^35^S]GTPγS binding to Gi/o(upper left), Gs/olf(upper right), and Gq/11(lower row) G-proteins stimulated by increasing concentrations (abscissa, log M) of agonists carbachol, furmethide oxotremorine and pilocarpine or antagonist N-methylscopolamine (NMS) is expressed as fold over basal (ordinate). Data are presented as mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equation 2 and results of fits are shown in the Table 4.
Kinetics of 500 nM [^3^H]GDP binding at individual subclasses of G-proteins measured in SPA is shown in Figure 9. Association of [^3^H]GDP with the Gi/o subclass of G-proteins that preferentially couple with the M2 receptors was biphasic (Figure 9, top left) with twice as many sites with fast (kobs1= 0.055 min^−1^) as with slow (kobs= 0.011 min^−1^) association kinetics (Table 5). Carbachol converted the association curve to become monophasic and decreased equilibrium binding 1.8-fold and ninefold at 10 and 100 µM concentrations respectively. Carbachol (100 µM) slowed down the association of [^3^H]GDP 12-fold in comparison to fast sites or 10-fold in comparison to mono-exponential fit of association data under control conditions (in the absence of carbachol) (kobs= 0.042 ± 0.005 min^−1^; Beq= 2.8 ± 0.3 fmol·µg·prot^−1^; mean ± SEM; n= 3). Dissociation curves were biphasic in the absence as well as in the presence of carbachol with 36 to 38% of slow binding sites. Carbachol accelerated the dissociation rate to a similar extent from both slow and fast sites. Acceleration was 6.3–6.5-fold by 10 µM carbachol and eightfold by 100 µM carbachol respectively (Figure 9, lower left; Table 5).
Figure 9 Kinetics of [^3^H]GDP binding to G
i/oand Gs/olfG-proteins. Association of 500 nM [^3^H]GDP with Gi/o(top left) and Gs/olf(top right) and dissociation of 500 nM [^3^H]GDP from Gi/o(bottom left) and Gs/olf(bottom right) subclasses of G-proteins was measured by scintillation proximity assay as described in Methods. GDP-less membranes were pre-incubated for 20 min with either buffer or 10 µM or 100 µM carbachol. Then, 500 nM [^3^H]GDP was added and association terminated by filtration at the indicated times (abscissa, min). [^3^H]GDP binding (ordinate) is expressed as fmol per µg of protein. In dissociation measurements, GDP-less membranes were equilibrated for two hours in the presence of 500 nM [^3^H]GDP. Dissociation was then initiated by the addition of 50 µM GDP alone or in combination with carbachol at 10 µM or 100 µM and terminated at indicated times (abscissa, min). [^3^H]GDP binding (ordinate) is expressed as per cent of binding at the beginning of dissociation. Data are expressed as mean ± SEM of values from three experiments performed in triplicate. Curves were fitted using Equations 5a and 5b (association) or 6a and 6b (dissociation). Results of fits are shown in Table 5.
Muscarinic M2 receptors also couple non-preferentially with the Gs/olf and Gq/11 subclasses of G-proteins. We therefore attempted to determine the influence of carbachol on the kinetics of [^3^H]GDP binding at these two other major G-protein subclasses. Unlike the results obtained for the Gi/o subclass, association and dissociation curves of [^3^H]GDP binding with Gs/olf were monophasic in the absence as well as in the presence of carbachol (Figure 9, right column). Carbachol had no effect on [^3^H]GDP binding association rate, accelerated [^3^H]GDP dissociation rate 1.8-fold and threefold, and decreased equilibrium binding 1.9-fold and 8.1-fold at 10 and 100 µM carbachol respectively (Figure 9, top right, Table 5). We were not able to determine the kinetics of [^3^H]GDP binding at Gq/11 subclass of G-proteins due to extremely fast on and off rates.
Effects of agonists on equilibrium binding of 10 µM [^3^H]GDP to individual subclasses of G-proteins was measured in SPA (Figure 10, Table 6). All agonists decreased [^3^H]GDP binding to Gi/o (Figure 10, upper left) and Gs/olf (Figure 10, upper right) G-proteins. Oxotremorine (Figure 10), unlike all other agonists, had no effect on [^3^H]GDP equilibrium binding to Gq/11 (Figure 10, lower panel) G-proteins. The rank order of factors of cooperativity between [^3^H]GDP and agonist binding varied among G-protein classes and was as furmethide = carbachol = oxotremorine > pilocarpine at Gi/o, carbachol = furmethide > oxotremorine > pilocarpine at Gs/olf, and carbachol > furmethide > pilocarpine > oxotremorine at Gq/11 (Table 6).
Figure 10 Allosteric interactions between agonists and [^3^H]GDP at individual G-proteins. The magnitude of allosteric interactions between agonists (carbachol, furmethide, oxotremorine, pilocarpine) and [^3^H]GDP was measured directly as changes in equilibrium binding of 10 µM [^3^H]GDP to G
i/o(upper left), Gs/olf(upper right) or Gq/11(lower row) G-proteins in the presence of increasing ligand concentration (abscissa, log M) via scintillation proximity assay as described in Methods. Data are presented as mean ± SEM of values from three experiments performed in quadruplicate. Curves were fitted using Equation 7. Results of fits are shown in the Table 6.
Conventional determination of agonist efficacy of G-protein coupled receptors often utilizes measurements of agonist-induced activation of GTPγS binding. We analysed the role of GDP (the second guanine nucleotide that binds to G-proteins) in the process of activation of the M2 muscarinic acetylcholine receptors and tested whether changes in its binding could serve as a possible measure of agonist efficacy. The muscarinic agonists studied here differ in structure as well as affinity and efficacy to stimulate GTPγS binding via the M2 muscarinic receptor (Figure 1). Binding studies show that GTPγS decreases the affinity of agonists as reported previously for the majority of, if not all, GPCRs (Wess, 1997). The decrease in agonist affinity is generally interpreted as being due to disintegration of the receptor/G-protein complex and the liberation of the signalling GTPγS-liganded G-protein α-subunit and complex of βγ subunits (Johnston and Siderovski, 2007). In accordance with previous findings (Haga et al., 1986; Florio and Sternweis, 1989; Tota and Schimerlik, 1990; Shiozaki and Haga, 1992) our data demonstrate that at the muscarinic M2 receptors GDP also decreases agonist affinity. In addition, we found that reduction of membrane-bound GDP increases the proportion of high-affinity binding sites for all agonists to a similar extent (Figure 2). Adding GDP back to GDP-less membranes reduces agonist affinity (Figure 7). These findings are consistent with the existence of an agonist low-affinity conformation of the receptor that is complexed with GDP-liganded G-protein, in addition to the agonist low-affinity conformation of receptor that is uncoupled from G-protein upon binding of GTP (Abdulaev et al., 2006).
Although the affinity of agonists at the low-affinity binding state is similar in the presence of either GDP or GTPγS, kinetics of guanine nucleotides binding provide evidence that the molecular mechanisms of modulation of agonist affinity is different. The ability of carbachol to accelerate dissociation and decelerate association of GDP (Figure 4) proves the existence of allosteric interaction between agonist and GDP on the receptor/G-protein complex. On the other hand, the inability of agonists to change the kinetics of GTPγS binding in the absence of GDP (Figure 3, left column; Table 3) is in concert with data obtained in a reconstituted system (Florio and Sternweis, 1989) and the commonly accepted concept that the GTPγS-liganded Gα subunit dissociates from receptor (Johnston and Siderovski, 2007) and therefore the kinetics of GTPγS binding cannot be allosterically regulated by agonists. Receptor-mediated acceleration of GTPγS association in the presence of GDP (Figure 3, upper row; Table 3) is a consequence of accelerated GDP dissociation, while in the absence of GDP the speed of GTPγS binding (irrespective of presence or absence of agonist) is already maximal. Lack of effect of agonists on the rate of GTPγS dissociation in both the presence and absence of GDP (Figure 3, lower row; Table 3) further supports the notion that, under our experimental conditions, the Gα subunit with bound GTPγS is not in physical contact with the receptor.
Agonist-induced allosteric acceleration of GDP dissociation from the Gα subunit strongly implies involvement of this mechanism in regulating the strength (efficacy) of agonist signal transmission to intracellular second messenger pathways. Despite multiple lines of evidence for allosteric interaction between agonist and GDP on receptor-G-protein complex the magnitude of these allosteric interactions has not been quantified so far. Our present data show that the magnitude of negative cooperativity between the four tested agonists displaying different potencies and efficacies, and GDP binding (Figure 6) demonstrate that full agonists (carbachol, furmethide) display significantly stronger negative cooperativity than partial agonists (oxotremorine, pilocarpine). The magnitude of negative cooperativity correlates with agonist efficacy in stimulating GTPγS binding to membranes (furmethide ≥ carbachol > oxotremorine > pilocarpine) (Figure 11). Interestingly, 30 years ago Birdsall et al. (1978) showed that agonist efficacy correlates with the ratio of agonist high- and low-affinity binding. Our results confirm these observations and provide a plausible interpretation. Agonist high-affinity binding takes place at a receptor-G-protein complex free of GDP and low-affinity binding occurs at a complex with GDP-liganded G-protein that is low due to negative cooperativity in binding of agonist and GDP. The stronger the negative cooperativity (more negative pα in our experiments) is, the higher the agonist efficacy and the lower the agonist affinity is in the low-affinity binding state. Thus, agonist efficacy correlates with the difference in affinities of the agonist high and low-affinity binding states.
Figure 11 Correlation of binding cooperativity between agonist and GDP with maximal stimulation of [^35^S]guanosine-5′-γ−thiotriphosphate (GTPγS) binding by agonist. Values of maximum of [^35^S]GTPγS binding (E
MAX) to membranes, Gi/o, Gs/olfand Gq/11(Tables 1 and 4) are plotted against the negative logarithms of the factor of binding cooperativity between agonist and GDP (pα) (results of Figure 6 and Table 6). Lines: 2D (Deming) linear regression of the data.
In addition to inhibition of adenylyl cyclase (Gi-mediated), activation of non-preferential G-proteins is associated with strong stimulation of adenylyl cyclase and relatively weak effects on accumulation of inositol phosphates (Gq-medited) by muscarinic M2 receptors was observed repeatedly (Ashkenazi et al., 1987; Burford et al., 1995; Jakubík et al., 1996; Michal et al., 2001; 2007;). In SPAs, furmethide, carbachol and pilocarpine stimulated GTPγS binding to preferential Gi/o as well as non-preferential Gs/olf and Gq/11 G-proteins. In contrast, oxotremorine stimulated GTPγS binding only to Gi/o and Gs/olf G-proteins (Figure 8, Table 4). Different orders of efficacies at individual G-protein classes can be explained by the concept of agonist specific conformations (Kenakin, 2003), where individual agonists induce different receptor conformations that differ in the ability to activate individual classes of G-proteins.
In agreement with an allosteric mode of action, affinities of GDP for the Gα subunits calculated from interactions with all of the tested agonists are the same (between 2.9 and 3.4 µM; Figures 6, 7 and 10) and correspond well to published values (Thomas et al., 1993) as well as results of [^3^H]GDP kinetics (Figures 4 and 9; Tables 3 and 5) and [^3^H]GDP saturation binding (Figure 5). Thus, changes in GDP affinity or kinetics are good measures of agonist efficacy at the Gi/o-coupled M2 muscarinic receptor. In practice, being the first step next to receptor activation, [^3^H]GDP binding appears to be a more direct measure of receptor activation than GTPγS binding or second messenger levels in case of M2 receptors, and this may be so at other Gi/o coupled GPCRs. However, this assay requires laborious preparation of membranes free of GDP. Agonist induced changes in GTPγS binding were demonstrated in fused Gsα/β2-adrenoceptors where agonist efficacy was well reflected by changes in the kinetics of GTPγS binding (Wenzel-Seifert and Seifert, 2000; Seifert et al., 2001). However, in concert with the involvement of agonist-induced decrease in GDP affinity in G-protein activation, GDP differentially and concentration-dependently influenced relative efficacies of partial agonists in increasing GTPγS binding (Wenzel-Seifert and Seifert, 2000). In accordance with previous findings (Florio and Sternweis, 1989), agonists at M2 receptors under our experimental conditions do not change the kinetics of GTPγS binding in the absence of GDP (Figure 3). Thus, while a change in the kinetic of GTPγS binding is a good measure of activation of physically coupled G-protein/β2- adrenoceptors, kinetics of GDP binding seem to be a closer measure in case of M2 muscarinic receptors and likely in other GPCR. Another drawback of GTPγS binding measurements is their dependence on the concentration of GDP that strongly affects outcome of the experiments (Figure 3). Also, unlike GDP binding, GTPγS concentration– response curve has to be measured under non-equilibrium conditions (Figure 3).
The data presented here show some interesting aspects of the process of receptor activation. NMS was reported as an inverse agonist at the M2 receptor (Jakubík et al., 1995; Burstein et al., 1997) and behaved as inverse agonist under our experimental conditions (Figures 1 and 8). Although positive cooperativity in binding with GDP would be expected, our data show that the cooperativity between NMS and GDP is neutral (Figures 6 and 7) and NMS only slightly slows down GDP dissociation (Figure 4 right), implying different mechanisms underlying the inverse agonist nature of NMS. One possible explanation may be that NMS stabilizes the receptor in the ground state (inactive conformation) (Hulme et al., 2003) that leads to reduction of spontaneous transition of the ligand-free receptor to an active state and slower rate of GDP dissociation. Significantly, this difference in mechanisms of agonism and inverse agonism cannot be revealed by measurement of GTPγS binding.
The second interesting aspect of our study is derived from data shown in Figure 9 that illustrates that carbachol slows down association of GDP with Gi/o G-proteins but does not change the rate of association of GDP with Gs/olf G-proteins. These data suggest that interaction of the M2 receptor with preferential Gi/o G-proteins differs from that with non-preferential Gs/olf G-proteins. One possible explanation is that Gi/o G-proteins precouple to M2 receptors while Gs/olf do not (Shea and Linderman, 1997; Hein et al., 2005), where precoupling gives an agonist a chance to influence GDP association while collision coupling does not. However, demonstration of this difference in coupling requires further detailed analysis. Again, this difference in kinetics at Gi/o and Gs/olf classes of G-proteins cannot be revealed by measurement of GTPγS binding.
In conclusion, we have demonstrated that the negative cooperativity between GDP and agonist binding played a key role in signal transduction via the M2 receptor. Agonist-induced low-affinity conformation of the Gα G-protein subunit for GDP leads to accelerated dissociation of bound GDP that in turn accelerates binding of GTP and G-protein activation. Thus, stronger negative cooperativity between a given agonist and GDP binding leads to a bigger shift of the GDP/GTP affinity ratio resulting in a higher rate of GTP binding and agonist efficacy. Our data demonstrated benefits of GDP binding measurements that can reveal mechanistic differences that are not apparent in measurements of GTP binding, as was demonstrated in case of inverse agonists versus agonists or Gi/o versus Gs/olf G-proteins. Measurements of GDP binding therefore provide additional information beyond that obtained from GTP binding measurements.
This work was supported by Project AV0Z 50110509, the grants of the Grant Agency of the Czech Republic 305/09/0681, the Grant Agency of the Czech Academy of Sciences IAA500110703, and the Ministry of Education, Youth and Sports, Czech Republic LC554, and NIH grant NS25743.
None.
Supporting Information: Teaching Materials; Figs 1–11 as PowerPoint slide.