Research ArticleGPCR SIGNALING

The mode of agonist binding to a G protein–coupled receptor switches the effect that voltage changes have on signaling

See allHide authors and affiliations

Science Signaling  03 Nov 2015:
Vol. 8, Issue 401, pp. ra110
DOI: 10.1126/scisignal.aac7419

Agonist control of GPCR voltage sensitivity

Most G protein–coupled receptors (GPCRs) are activated by ligand binding, but some are also affected by changes in plasma membrane potential, which can either enhance or inhibit GPCR-mediated signaling. Through FRET-based experiments in single cells, Rinne et al. found that depolarization enhanced signaling by the M3 muscarinic acetylcholine receptor when the receptor was bound to the agonists choline or pilocarpine; however, depolarization attenuated M3 receptor signaling when either carbachol or acetylcholine was bound. Molecular docking simulations showed that each group of agonists adopted a distinct binding position. Mutation of a critical residue in the binding pocket changed the binding position of carbachol and switched the response of the carbachol-bound receptor so that signaling was enhanced by membrane depolarization. Together, these data suggest that the binding mode of the agonist determines whether membrane potential changes will enhance or attenuate GPCR signals. These results provide a potential molecular mechanism for drugs that are agonists of specific GPCRs, yet have distinct effects.


Signaling by many heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptors (GPCRs) is either enhanced or attenuated by changes in plasma membrane potential. To identify structural correlates of the voltage sensitivity of GPCR signaling, we chose muscarinic acetylcholine receptors (the M1, M3, and M5 isoforms) as a model system. We combined molecular docking analysis with Förster resonance energy transfer (FRET)–based assays that monitored receptor activity under voltage clamp conditions. When human embryonic kidney (HEK) 293 cells expressing the individual receptors were stimulated with the agonist carbachol, membrane depolarization enhanced signaling by the M1 receptor but attenuated signaling by the M3 and M5 receptors. Furthermore, whether membrane depolarization enhanced or inhibited receptor signaling depended on the type of agonist. Membrane depolarization attenuated M3 receptor signaling when the receptor was bound to carbachol or acetylcholine, whereas depolarization enhanced signaling when the receptor was bound to either choline or pilocarpine. Docking calculations predicted that there were two distinct binding modes for these ligands, which were associated with the effect of depolarization on receptor function. From these calculations, we identified a residue in the M3 receptor that, when mutated, would alter the binding mode of carbachol to resemble that of pilocarpine in silico. Introduction of this mutated M3 receptor into cells confirmed that the membrane depolarization enhanced, rather than attenuated, signaling by the carbachol-bound receptor. Together, these data suggest that the directionality of the voltage sensitivity of GPCR signaling is defined by the specific binding mode of each ligand to the receptor.


Muscarinic receptors for acetylcholine represent a family of heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptors (GPCRs) that are found in many cell types, including electrically excitable cells. Five receptor isoforms (M1 to M5) are known, which signal either through Gq proteins (for M1, M3, and M5) or through Gi/o proteins (for M2 and M4). Depending on the receptor isoform and its tissue distribution, muscarinic acetylcholine receptors control various physiological functions and represent important pharmacological targets (1). Studies of M1 and M2 receptors established that GPCR signaling is sensitive to the plasma membrane potential (2). A depolarization of the membrane causes either activation of M1 receptors or deactivation of M2 receptors. In response to membrane depolarization, both receptor subtypes display voltage-dependent gating currents with properties similar to the gating currents of ion channels (26). Voltage-dependent effects on signaling have been further detected in P2Y1 (7, 8), lysophosphatidic acid (9), glutamate (10), dopamine (11), histamine (12), and α2A adrenergic receptors (13). Nevertheless, a unifying molecular mechanism of how GPCRs sense voltage is still elusive. Parameters that influence the precise nature of the voltage-dependent effects on signaling have been suggested to include (i) the charge of the agonist, (ii) the G protein–GPCR interaction, and (iii) the G protein class itself; for example, depolarization appears to deactivate Gi-coupled GPCRs, whereas it activates Gq-coupled GPCRs (1416).

Here, we compared voltage-dependent effects on the signaling of the Gq-coupled receptors M1, M3, and M5. We resolved receptor function and downstream signaling in relation to the membrane potential (VM) with biosensors that relied on Förster resonance energy transfer (FRET). All experiments were performed with single human embryonic kidney (HEK) 293 cells, which enabled precise control of the membrane potential by means of voltage clamp. The crystal structure of the M3 receptor bound to the ligand tiotropium (17) enabled us to perform docking calculations to place the agonists acetylcholine, carbachol, choline, and pilocarpine in the orthosteric binding pocket of the M3 receptor. The resulting binding poses were specific for each ligand, with two binding modes that were distinguishable from each other. A mutational analysis of the orthosteric binding site of the M3 receptor in silico and in live cell experiments demonstrated that the specific ligand binding pose, but not coupling to G proteins, defined how voltage regulated receptor function. These results shed light on a putative allosteric molecular mechanism for the voltage sensitivity of GPCRs.


The active conformation of the M1 receptor is voltage-dependent

Activation of the M1 receptor by carbachol was quantified with the FRET-based receptor biosensor M1-cam (18) in HEK 293 cells under whole-cell voltage clamp conditions. The biosensor reported receptor activation by rapid attenuation of an intramolecular FRET signal generated between cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP) variants inserted in the receptor molecule (Fig. 1A). Application of carbachol induced a switch of the biosensor into an active conformation, which was detected as a rapid decrease in the ratio F535/F480 (termed FRET hereafter; Fig. 1B, top). Depolarization of the plasma membrane from −90 to +60 mV caused more activation of M1-cam (Fig. 1B, bottom). The same change in voltage in the absence of agonist had no effect on the conformation of the receptor biosensor (Fig. 1B, bottom). Depolarization caused a shift in the concentration-response relationship for carbachol: M1-cam was three times more sensitive to carbachol at +60 mV than it was at −90 mV (Fig. 1C).

Fig. 1 The M1 receptor is sensitive to voltage changes.

(A) Scheme of the FRET-based receptor biosensor M1-cam. Ligand binding induces a conformational change in the receptor that results in decreased FRET. (B) Activation of M1-cam in HEK 293 cells by carbachol (CCh) was detected as a decrease in FRET (F535/F480). Upon application of carbachol, M1-cam displayed robust activation at −90 mV and switched to an even more active state during a depolarization step to +60 mV (top panel, n = 6 cells). The biosensor was not activated by voltage alone (bottom panel, n = 5 cells). (C) Concentration-response curves for the carbachol-dependent activation of M1-cam in HEK 293 cells at −90 mV (black circles) and at +60 mV (open circles). Data are from paired experiments and are means ± SEM of 11 individual cells per data point. (D) FRET response of M1-cam in HEK 293 cells to 10 μM carbachol at different membrane potentials in the presence (black curve, solid circles) or absence (blue curve, open circles) of 500 μM intracellular GTPγS. The data were normalized to the holding potential (FRET/FRET−90 mV) and fitted to a Boltzmann equation. The calculated V0.5 values of M1-cam for control cells and GTPγS-treated cells are depicted. Data are from 11 to 14 cells per data point. The detailed experimental protocol is shown in fig. S1. All data are means ± SEM of the indicated number of individual cells.

Previous studies showed that the M2 receptor displays voltage dependence very selectively in a high-affinity state, which is induced by intracellular docking to Gi/o proteins (3, 19, 20). We were interested in whether this mechanism held true for the closely related M1 receptor and its binding partner Gq. To address this, we loaded HEK 293 cells transfected with plasmid encoding M1-cam with guanosine 5′-O-(3′-thiotriphosphate) (GTPγS) (500 μM) through the patch pipette. As a nonhydrolyzable analog of guanosine 5′-triphosphate (GTP), GTPγS promoted an irreversible activation of Gq in response to carbachol, thereby permanently uncoupling the G protein from the receptor (fig. S1). Cells loaded with GTPγS were subjected to various membrane potentials in the presence of carbachol (see fig. S1 for details). The response of the biosensor to carbachol was normalized to the FRET amplitude at −90 mV (FRET/FRET−90 mV), plotted against VM, and fitted to a Boltzmann equation to obtain voltage-activation relationships (Fig. 1D). The curves with and without GTPγS were indistinguishable from one another, which showed that M1-cam was activated by depolarization in an identical manner in its G protein–free state (+GTPγS, blue curve: V0.5 = −46 mV) compared to control cells (−GTPγS, black curve: V0.5 = −53 mV). These data suggest that for the M1 receptor, the agonist-receptor complex (that is, the active conformation of the receptor) is voltage-sensitive independently of its coupling to G proteins.

Depolarization activates signaling downstream of the M1 receptor

We also explored whether voltage-induced changes in receptor activation modulated cellular signaling. We assessed the extent of Gq protein activation with a FRET-based biosensor that monitors the kinetics of the Gq protein cycle (Fig. 2A) (21, 22). Application of carbachol resulted in a concentration-dependent decrease in the FRET signal, which reflected the dissociation of the G protein subunits during activation (Fig. 2B). Under voltage clamp conditions, the Gq protein was more active at +60 mV (Fig. 2C, red traces) than at −90 mV (Fig. 2C, black traces) in transfected cells expressing the M1 receptor. The canonical cellular response to stimulation of Gq-coupled GPCRs is the inositol 1,4,5-trisphosphate (IP3)–induced release of Ca2+ from internal stores (23). In HEK 293 cells transfected with plasmids encoding the M1 receptor and the Ca2+ indicator protein Green-GECO1.1 (24), a typical Ca2+ transient was induced by 50 nM carbachol, but no Ca2+ signal was detected at lower concentrations of agonist (Fig. 2D). The corresponding extent of activation of the Gq protein in response to 50 nM carbachol was about 25% of the maximal G protein activation (Fig. 2B). Because the potentiation of Gq activation at +60 mV was in a similar range (Fig. 2C), we tested whether the voltage-induced potentiation of Gq activation stimulated intracellular Ca2+ release. Indeed, in the presence of 20 nM carbachol, a voltage step to +60 mV stimulated robust Ca2+ release from intracellular Ca2+ stores (Fig. 2E). A comparable Ca2+ transient was never observed at −90 mV (Fig. 2F) or during depolarization of the cells in the absence of carbachol (Fig. 2G).

Fig. 2 Signaling of the M1 receptor to Gq is sensitive to voltage changes.

(A) Scheme of a FRET biosensor that monitors the Gq protein cycle. (B) In transfected HEK 293 cells expressing the M1 receptor and the Gq sensor, 50 nM carbachol induced Gq activation at about 25% of the maximal extent, which was induced by 10 μM carbachol. The trace represents mean ± SEM of five individual cells. (C) Average traces of Gq sensor activation in HEK 293 cells in response to carbachol (left) or pilocarpine (right) at −90 mV (black) or +60 mV (red). The traces are normalized to the maximal activation of the Gq sensor (induced by 10 μM carbachol). Data are average traces of four to six cells per membrane potential and per agonist. (D) Representative trace of a global Ca2+ transient of a HEK 293 cell transfected with plasmid encoding the Ca2+ indicator Green-GECO1.1. 50 nM carbachol induced a detectable Ca2+ transient in six of eight cells tested. (E) Representative Ca2+ transient of a cell that was exposed to 20 nM carbachol and then subjected to a voltage step from −90 mV to +60 mV. Four of 10 cells responded to the voltage switch with a Ca2+ signal. (F and G) No Ca2+ transients were detected in the presence of 20 nM carbachol at −90 mV (F; six cells tested) or at +60 mV in the absence of carbachol (G; five cells tested).

The effect of voltage on receptor function depends on the receptor subtype

The Gq protein biosensor enabled us to compare the voltage sensitivities of wild-type M1, M3, and M5 receptors. Cells were transfected with plasmid encoding the M1 receptor and stimulated with 100 nM carbachol. When subjected to a dynamic voltage clamp protocol, depolarization to +60 mV caused reversible potentiation of Gq activation (Fig. 3A). In contrast, cells transfected with plasmids encoding either the M3 receptor (Fig. 3B) or the M5 receptor (fig. S2) responded to the voltage step with reduced Gq activation. This effect was evident for the M3 receptor over a range of carbachol concentrations, and the M3 receptor was only half as sensitive to carbachol at +60 mV as it was at −90 mV (Fig. 3C). As an alternative signaling pathway, we tested arrestin signaling. Arrestins recognize GPCRs in an active, phosphorylated configuration, which is free of G proteins (25). The receptor-arrestin interaction was inferred by detection of an increase in FRET between turquoise-labeled arrestin3 and YFP-tagged receptors (Fig. 3D) (26). Application of 1 μM acetylcholine to receptor-expressing cells at −90 mV resulted in a rapid increase in FRET amplitude, which suggested that arrestin was recruited to activated M1 (Fig. 3E) and M3 (Fig. 3F) receptors. During depolarization to +60 mV, we observed a rapid recruitment of additional arrestin to the M1 receptor, whereas we saw dissociation of arrestin from the M3 receptor, which confirmed the opposing behaviors of the two receptors in response to depolarization. Together, these data provide evidence that the membrane potential regulates signaling downstream of muscarinic acetylcholine receptors with a directionality that varies between the specific receptor subtypes.

Fig. 3 The effect of depolarization on Gq-coupled muscarinic acetylcholine receptors is receptor subtype–specific.

(A and B) The activity of wild-type (WT) M1 (A) and M3 (B) receptors was assessed with the Gq FRET biosensor. Application of carbachol resulted in activation of both receptors at −90 mV. During depolarization of the membrane to +60 mV, the M1 receptor induced increased activity of Gq (A; average trace of seven individual cells), whereas the activity of Gq during depolarization was reduced in cells expressing the M3 receptor (B; average trace of six individual cells). (C) Determination of M3 receptor sensitivity to carbachol at +60 mV and at −90 mV. Data are from five to seven cells per concentration tested. (D) Scheme of an arrestin FRET biosensor. Recruitment of arrestins by the active GPCR is detected as an increase in the FRET ratio. Tur, turquoise. (E) Activation of the M1 receptor by acetylcholine (ACh) led to an enhanced FRET signal at −90 mV that was facilitated by depolarization of the membrane (average trace of four individual cells). (F) Activation of the M3 receptor by acetylcholine resulted in a similar FRET signal at −90 mV, which was reduced during depolarization of the membrane to +60 mV (average trace of eight individual cells). All average traces are presented as means (solid lines) ± SEM (gray shading) of the indicated number of individual cells.

The binding mode of agonists affects the voltage dependence of the M3 receptor

Muscarinic receptor activation was voltage-sensitive only in the presence of agonist (Fig. 1B). We tested the voltage dependence of carbachol- and pilocarpine-activated M3 receptors and found a specific ligand-induced response. In the case of carbachol, M3 receptor activation was reduced at positive membrane potentials, whereas in the case of pilocarpine, positive membrane potentials resulted in potentiated receptor activation (Fig. 4, A and B). This phenomenon has also been observed for the M2 receptor (6). Therefore, we investigated the ways in which the molecular interaction of carbachol with the receptor might differ from that of pilocarpine. In general, ligands of class A GPCRs interact with the receptor at a conserved hydrophobic binding pocket, which is formed by the transmembrane domains (27). We performed multiple molecular docking calculations to identify key amino acid residues that directly interacted with ligands (28). To enable accurate docking calculations that resulted in reliable binding poses of the ligands, we focused on the M3 receptor rather than on the other two receptor isoforms because of the availability of the crystal structure of the M3 receptor with a ligand bound to the orthosteric site [Protein Data Bank (PDB) ID: 4DAJ] (17). In silico, we docked acetylcholine, carbachol, choline, or pilocarpine to the binding pocket of the M3 receptor using the crystal structure with its cocrystallized ligand removed. For the wild-type receptor, the binding poses of carbachol (Fig. 4C) and acetylcholine (Fig. 4D) were very similar and were mainly characterized by a hydrogen bond interaction with the side chain of asparagine 6.52 (N6.52) localized to transmembrane helix 6 (TM6), which caused both ligands to adopt an orientation toward this helix. In experiments with live cells, the M3 receptor was deactivated by depolarization to +60 mV in the presence of either acetylcholine or carbachol (Fig. 4E). Although the binding poses for both agonists were very similar, we observed that the degree of voltage-dependent deactivation varied and appeared much less pronounced for carbachol than for acetylcholine (Fig. 4E). In contrast, the predicted binding mode of pilocarpine (Fig. 4F) was different, favoring an orientation more toward aspartate 3.32 of TM3 (D3.32), which is more distant from TM6, and having no discernible interaction with N6.52. TM6 was identified previously as a critical structural element that is involved in the transition of receptors from inactive to active conformations (27). Therefore, we hypothesized that these different agonist binding modes caused the opposite responses to depolarization by carbachol- and pilocarpine-activated wild-type M3 receptors.

Fig. 4 Sensitivity of the M3 receptor to voltage changes is ligand-specific.

(A) Average traces of Gq activation in HEK 293 cells expressing the M3 receptor and the Gq biosensor were measured in response to carbachol (left) and pilocarpine (right) at −90 mV (black) and +60 mV (red). (B) Representative trace of a single cell at −90 mV that was exposed to 100 nM carbachol or to 1 μM pilocarpine and subjected to depolarization of the membrane to +60 mV during the application of each agonist. Note that carbachol was removed completely before pilocarpine was applied (dotted line). (C and D) Calculated binding poses of carbachol and acetylcholine docked to the orthosteric binding pocket of the WT M3 receptor. The poses of carbachol (C) and acetylcholine (D) were very similar and showed an interaction with N6.52 of TM6. (E) Voltage-induced attenuation of receptor activation for 100 nM carbachol and 10 nM acetylcholine. The trace represents an average of four single transfected HEK cells expressing the M3 receptor and the Gq protein biosensor. Carbachol was removed completely before acetylcholine was applied, as indicated by the dotted line. (F) The pose of pilocarpine showed a different orientation toward D3.32 of TM3. The black dashed lines highlight ligand-receptor interactions with residues 3.32 or 6.52 of the orthosteric site. The black numbers indicate minimum distances (in angstrom) between the ligand and the interacting residue. See table S1 for details.

To identify which amino acid in the orthosteric binding pocket of the M3 receptor had a major influence on the binding mode of agonists, we performed a series of in silico mutations of six active site residues that are in contact with the ligand, and we performed the subsequent docking calculations. Among all of the residues that were mutated computationally, N6.52 of TM6 was identified as the key residue for the preservation of the native binding modes for carbachol (Fig. 4C) and acetylcholine (Fig. 4D); thus, the substitutions N6.52Q and N6.52A were analyzed in more detail. A glutamine at position 6.52 added a slightly bulkier side chain to this residue and resulted in an increase in the distance between carbachol and TM6, with an effective loss of the hydrogen bond to residue 6.52 (Fig. 5A and table S1). The poses for each ligand were also stable: During repetitive docking calculations, we changed the number used to start the random number generator in our docking program for each calculation, which could affect the docking results. Even though we observed small differences in the resulting binding poses (fig. S3), the overall binding modes were similar. Thus, docking in silico to the mutated receptor predicted an alternative “pilocarpine-like” binding mode for carbachol, which was oriented toward TM3 (Fig. 5A), whereas the binding poses for acetylcholine (Fig. 5B) and pilocarpine (Fig. 5C) did not change.

Fig. 5 Analysis of ligand binding and the effects of voltage changes on signaling by the N6.52Q mutant M3 receptor.

(A to C) Calculated binding poses for carbachol (A), acetylcholine (B), and pilocarpine (C) docked to the N6.52Q mutant M3 receptor. The side-chain mutation N6.52Q (green) induced a previously uncharacterized binding pose for carbachol (A) toward TM3, with fewer interactions with TM6. In contrast, the binding poses of acetylcholine (B) and pilocarpine (C) appeared to be similar to those for the WT receptor (compare to Fig. 4). The black lines highlight ligand-receptor interactions with residues 3.32 or 6.52 of the orthosteric site. The black numbers indicate minimum distances (in angstrom) between the ligand and the interacting residue. See table S1 for details. (D) The N6.52Q mutation reversed the directionality of the effect of a voltage change on the signaling of the mutant M3 receptor in response to carbachol (1 μM carbachol, average trace of five individual cells in red) compared to the effect of a voltage change on the WT receptor (50 nM carbachol, black average trace from Fig. 3B). The control N6.52A mutation did not change the effect of a voltage change on receptor signaling (1 μM carbachol, average trace of four individual cells, binding pose in fig. S4). (E) The N6.52Q mutation did not change the effect of voltage change on receptor signaling induced by 100 nM acetylcholine (average trace of five individual cells). Average traces are means (solid trace) or as means (solid trace) ± SEM (gray shading).

In subsequent cellular experiments, the polarity of the voltage dependence of the N6.52Q mutant M3 receptor was reversed in the presence of carbachol (Fig. 5D, red trace) compared to that of the wild-type receptor (Fig. 5D, black trace). In contrast, the substitution of alanine for asparagine at position 6.52 (N6.52A) added a smaller side chain to this residue (fig. S4), and this mutant displayed a voltage dependence similar to that of the wild-type M3 receptor (Fig. 5D, blue trace). This direct relation between agonist binding pose and whether depolarization enhanced or attenuated receptor signaling was confirmed in experiments in which acetylcholine and choline were used as agonists. In the N6.52Q mutant M3 receptor, the binding poses of acetylcholine (Fig. 5B) and choline (fig. S5) did not differ from their corresponding docking poses in the wild-type M3 receptor (Figs. 4 and 5 and fig. S5). Consistent with this finding, the effect of depolarization on receptor signaling in response to acetylcholine (Fig. 5E) or choline (fig. S5) was not altered by this mutation. We did not test the effect of depolarization on pilocarpine-activated M3 receptor variants because we observed a pronounced loss in the affinity of M3 N6.52Q for pilocarpine and no substantially altered docking pose of pilocarpine in the receptor mutant (Fig. 5C). Because N6.52 is conserved among the muscarinic acetylcholine receptor family, we also introduced the N6.52Q mutation to the M1 and M5 receptors and tested their voltage sensitivities. The response of the N6.52Q mutant M1 receptor to depolarization was similar to that of the wild-type M1 receptor (compare fig. S6 to Fig. 3), which suggests that carbachol bound to the M1 receptor in a manner similar to that of pilocarpine, with no detectable influence of residue 6.52 of TM6. In contrast, voltage did not change the activity of the N6.52Q mutant M5 receptor in the case of carbachol but caused a reduction in signaling activity when acetylcholine was used (deactivation, fig. S2). Thus, the N6.52Q mutant M5 receptor resembled to a certain degree the N6.52Q mutant M3 receptor in terms of voltage sensitivity.

An alternative approach to underscore these findings would benefit from side-chain mutations of the “choline site” of the M3 receptor to alter agonist binding in a different region of the binding pocket. Docking calculations (fig. S4) showed that the D3.32E mutation in the M3 receptor resulted in an elongation of the side chain, which appeared not to affect the binding poses for carbachol or pilocarpine; however, the D3.32E mutation resulted in barely functional receptors in cellular experiments, even in the very sensitive Gq protein assay (fig. S4). Note that all mutations of the orthosteric binding site used here affected agonist binding. This phenomenon did not occur because of low receptor abundance or the intracellular localization of misfolded receptors (fig. S7) but rather was attributed to changes in agonist affinity. For carbachol and acetylcholine, we measured a 10-fold reduction in their affinities for the N6.52Q and N6.52A mutant M3 receptors in comparison to their affinity for the wild-type M3 receptor (figs. S4 to S6). This effect was compensated for during our experiments by adjusting the absolute agonist concentration to reach comparable degrees of receptor activation between wild-type and mutant receptors. For pilocarpine and choline, the shift in affinity was even greater, and we were unable to measure complete concentration-response curves because of a lack of saturation at the highest feasible concentrations.

The prediction from the docking calculations that amino acid 6.52 of TM6 had a substantial effect on the agonist binding mode and on receptor activation is consistent with TM6 being a critical structural element that undergoes a large movement during receptor activation. In addition, N6.52 forms hydrogen bonds with the ligand tiotropium in the crystal structure of the M3 receptor (17). We compared the position of N6.52 within the core of active and inactive receptors with the structures of the active form of the M2 receptor (PDB ID: 4MQS and 4MQT) and with the structures of the inactive forms of the M3 (4DAJ) and M2 (3UON) receptors (Fig. 6A). We found that N6.52 was localized in the flexible outer membrane region of TM6, and we followed its movement during activation into the receptor core (Fig. 6A, arrows), presumably maintaining the ligand-TM6 interaction. Furthermore, we found no evidence that the N6.52Q mutation hindered this movement (Fig. 6B). In conclusion, our data provide evidence that a particular agonist binding mode induces a specific receptor conformation and determines whether a change in membrane voltage enhances or attenuates signaling by the M3 receptor.

Fig. 6 Movement of residue 6.52 in TM6 during receptor activation in the WT and N6.52Q mutant M3 receptors.

(A) Side view (left) and top view (right) of the receptor core highlighting TM5 and TM6. Left: The side-chain conformation of 6.52 in the structure of the inactive M2 receptor (3UON, light green) is very similar to the conformation of the inactive M3 receptor (4DAJ, dark green). The outer bilayer segment of helix 6 and N6.52 are closer to the receptor core in the structures of the active M2 receptor (4MQS in yellow and 4QMT in purple). Right: The different locations of TM6 for both activation states as viewed from the top. The arrows suggest the movement of N6.52 during receptor activation. (B) Superposition of the structures of inactive M3 receptor, its N6.52 mutant, and the active M2 receptor. The inactive state of the WT M3 receptor (green) is similar to the inactive state of the N6.52Q M3 mutant receptor, as shown in salmon. The active state of the M2 receptor is depicted in yellow. The elongated side chain of the N6.52Q mutant M3 receptor does not interfere with receptor activation.


Here, we compared voltage-dependent effects on the signaling of the Gq-coupled M1, M3, and M5 receptors with FRET-based assays that assessed conformational changes in the receptor or of downstream effectors. These direct approaches provided several observations regarding the voltage control of the muscarinic acetylcholine receptors. First, they showed that the agonist-receptor complex itself was voltage-sensitive independently of Gq proteins, in contrast to the mechanism previously proposed for the voltage sensitivity of the M1 or M2 receptors (2, 19, 20). Second, these experiments showed that each particular binding pose of an agonist induced a specific receptor conformation that defined the polarity of the voltage-dependent effect on signaling.

We characterized the voltage-dependent effects on the signaling of Gq-coupled muscarinic GPCRs at the level of the receptor with a biosensor. FRET-based receptor biosensors are useful for this approach because they provide a direct tool to relate conformational changes of the protein with the degree of receptor activation, which is closely related to the amount of agonist bound to the receptor (4, 2931). We found that the function of the M1 receptor was potentiated by depolarization (Fig. 1). Previous studies of the M2 receptor suggested that voltage sensitivity exists only in the high-affinity state of the receptor, which is maintained by intracellular coupling to Gi/o proteins. Consequently, the voltage dependence of M2 receptor (M2-ACh-R) was abolished upon treatment with pertussis toxin (PTX) (2, 19), which prevents GPCRs from coupling to Gi/o proteins. In the case of the closely related M1 receptor, however, our studies showed that voltage dependence was independent of the interaction between the receptor and the Gq protein; loading of the cells with GTPγS at a concentration that was sufficient to irreversibly activate Gq, a procedure that uncouples the receptor from the Gq protein cycle, did not abolish the voltage sensitivity of the receptor (Fig. 1D and fig. S1). Our results are consistent with the findings of another study, which reported that neither removal of G protein interactions (by treatment with GTPγS) nor overexpression of Gq led to robust changes in the voltage dependence of M1 receptors (4). Furthermore, in our own experiments, M1 and M3 receptors were voltage-sensitive when occupied by arrestin (Fig. 3), which is considered to be a G protein–free state (25). Moreover, we previously found that the α2A adrenergic receptor, which couples to Gi/o similarly to muscarinic acetylcholine receptors, also proved to be voltage-sensitive after treatment with PTX or GTPγS (13). Together, these observations indicate that the voltage sensitivity of GPCRs does not require coupling to G proteins. Accordingly, whether the effect of voltage sensitivity on GPCR signaling is positive or negative is also not determined by the class of G protein to which the receptor couples.

Intracellular signaling stimulated by the M1 receptor was voltage-dependent. Depolarization of the membrane potential over a physiological range enhanced the activation of Gq proteins (Fig. 2D), sensitized Ca2+ signaling (Fig. 2G), and potentiated arrestin signaling (Fig. 3E). Potentiation of Gq-coupled GPCR signaling by depolarization was described previously for the M1 receptor (4) and P2Y1 receptors (7, 8); however, our study reports that membrane depolarization attenuates the function of some Gq-coupled GPCRs: the M3 and M5 receptors displayed reduced activity at positive membrane potentials at the levels of G protein signaling (Fig. 3B and fig. S2) and arrestin signaling (Fig. 3F). We conclude that neither the charge of the agonist carbachol nor the class of G proteins determined the directionality of the voltage dependence of the receptors used in this study.

For the M3 receptor, whether depolarization had a positive or negative effect on signaling was directly influenced by the structure and corresponding molecular binding pose of the ligand: Agonists that bound with an orientation to the “acetyl side” of the binding pocket (acetylcholine and carbachol) reached a binding mode that was oriented toward TM6 because of a polar interaction with residue N6.52 (Fig. 4). This binding mode was associated with reduced receptor activity in response to membrane depolarization. On the other hand, agonists such as choline and pilocarpine reached an orientation toward D3.32 of TM3 or the “choline side” of the binding pocket (Fig. 4 and fig. S5). This binding mode was associated with enhanced receptor activation upon membrane depolarization. Consequently, the N6.52Q mutation in the M3 receptor, which caused carbachol to bind to the choline side of the binding pocket in a pilocarpine-like manner (Fig. 5), reversed the response of the receptor to depolarization. These binding pose–specific responses to membrane depolarization were comparable to the N6.52Q mutant M5 receptor because deactivation in response to depolarization, as was observed for the wild-type M5 receptor (fig. S2), was absent for carbachol-activated receptor mutants but was well pronounced when acetylcholine was used as an agonist (fig. S2). On the contrary, the N6.52Q mutation did not alter the response of the M1 receptor to depolarization, suggesting that carbachol binds to wild-type M1 receptors already in a pilocarpine-like manner.

The next consequent step would be a mutation experiment to alter pilocarpine binding in M3 receptors into a “carbachol-like” binding mode by extending the amino acid side chain at position D3.32 (Fig. 4F). This experiment failed because this particular mutation abolished the functional response of the mutant receptors (fig. S4). The latter was not attributed to changes in affinity or ligand binding but to a 100-fold decrease in efficacy (32). This is consistent with our docking calculations, which predicted the normal binding of all agonists to the D3.32E mutant. Although we can only speculate about the specific effect of voltage on N6.52Q-mutated receptors, our results favor the hypothesis that membrane depolarization induces a more unfavorable conformation of the “acetyl binding” region of the orthosteric binding site, whereas it leads to a more favorable conformation of the “choline” binding pocket. Whether a ligand exhibits an increased or decreased binding affinity toward a receptor under depolarized conditions depends on the relative contribution of these two binding sites to the total binding affinity of the ligand. Such a voltage-induced movement of the binding pocket itself might also explain how a ligand that normally serves as an antagonist is converted to an activator of the receptor during depolarization of the membrane (8).

Collectively, the mutational studies showed that (i) carbachol binds to the orthosteric binding site of the M1 receptor in a distinct way that is similar to that of pilocarpine, which is different from its binding mode in the M3 and M5 receptors, and (ii) each particular binding pose determines whether depolarization has a stimulatory or inhibitory effect on receptor signaling. Note that our study did not identify a particular molecular structure of the muscarinic acetylcholine receptors that served as an intrinsic voltage sensor. A putative mechanism for the voltage sensitivity of GPCRs may differ substantially from the known mechanisms of ion channels. The latter contain several positively charged amino acids that are concentrated in the S4 segment of the protein, causing a detectable movement of more than four charges through the electrical field across the membrane (14). GPCRs do not contain such a motif, and presumably a maximum of one charge moves in a muscarinic acetylcholine receptor in response to membrane depolarization because the calculated z values are below or close to 1 (3, 6). This is consistent with the voltage dependence of agonist-bound M1-cam in this study (z = 0.76). The physiological relevance of voltage-sensitive Gq-coupled GPCR signaling may lie in the fine-tuning of Ca2+ signals at low concentrations of agonist (Fig. 2), a phenomenon that has been observed in a similar context for muscarinic receptors in a neuronal cell line (33) and for purinergic receptors in megakaryocytes (8). In both cases, voltage-enhanced activation of Gq-coupled GPCRs caused potentiation of Gq-mediated, phospholipase C (PLC)– and IP3-dependent Ca2+ release from internal stores. The physiologically relevant, half-maximal voltage that we observed for the activation of M1 receptors by intramolecular FRET (−53 mV) supports this conclusion. In addition, under conditions in which depolarization potentiates Gq-coupled receptors, it may serve as a mechanism of coincidence detection to integrate electrical and chemical signaling events, particularly at low or subthreshold concentrations of agonist (Fig. 2E) (8). In conclusion, we proposed that voltage sensitivity represents a mechanism to fine-tune GPCR signaling, which depends on ligand-receptor interactions. Therefore, the methods demonstrated here may prove to be useful tools to better understand the dynamics and functions of GPCRs in general.


Cell culture and transfections

All experiments were performed with stably or transiently transfected HEK 293 cells. Cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 1% glutamine, 10% fetal calf serum, and penicillin and streptomycin using standard cell culture conditions. To analyze the M1-cam biosensor (18), a stable cell line was generated by transfecting HEK 293 cells with 1 μg of M1-cam complementary DNA (cDNA) and subsequently culturing the cells under selection with G418 (29). To monitor Gq protein activation, HEK 293 cells were transfected with the following cDNAs (amounts given as μg cDNA per 6-cm culture dish): human M1, M3, or M5 receptor (0.5), murine Gαq-YFP (1.0), human Gβ1-Cerulean (0.5), human Gγ2 (0.2), and human G protein–coupled receptor kinase 2 (GRK2) (0.5); to measure intracellular Ca2+ flux: M1 or M3 receptor (0.5) and CMV-Green-GECO1.1 (1 μg), which was provided by R. Campbell through Addgene (plasmid #32445) (24); to analyze interactions between GPCRs and Gβ: M3-YFP or M3-YFP (N6.52Q) (0.5), Gβ1-Cerulean (0.5), murine Gαq (0.8), and Gγ2 (0.2); to monitor the recruitment of arrestin3: human M1 or M3 receptor (0.6), murine turquoise-arrestin3 (0.6), and human GRK2 (0.8). All mutations were introduced into the human cDNAs encoding muscarinic acetylcholine receptors by polymerase chain reaction and were verified by sequencing. The following mutagenesis primers were used: M1-ACh-R (N6.52Q), 5′-acctggacaccgtaccaaatcatggtgctggtg-3′; M3-ACh-R (N6.52Q), 5′-acctggacccctaccaaatcatggtgctggtg-3′; M5-ACh-R (N6.52Q), 5′-cacatggaccccgtatcagatcatggtcctggtttc-3′; M3-ACh-R(N6.52A), 5′-acctggaccccatacgccatcatggttctggtg-3′; M3-ACh-R (D3.32E), 5′-ctctggcttgccattgagtacgtagccagc-aatgc-3′. All constructs were subcloned into the plasmids pcDNA1 or pcDNA3 (Invitrogen). Cells were transfected with Effectene reagent (Qiagen) according to the manufacturer’s instructions, seeded on sterile, poly-l-lysine–coated glass coverslips. Fluorescence measurements were performed 48 hours after transfections.

Fluorescence microscopy and electrophysiology

All experiments were performed with single HEK 293 cells subjected to voltage clamp in the whole-cell configuration. Fluorescence was recorded with an inverted microscope (Zeiss Axiovert 135) equipped with an oil immersion objective (Zeiss A-Plan 100×/1.25), a Polychrome V illumination source (TILL Photonics), a dual emission photometry system (TILL Photonics), and a hardware and software package (EPC10 amplifier with digitization interface, Patchmaster software version 5.52, HEKA). For FRET measurements, cells were excited at 435 nm, and the corresponding emitted fluorescence from CFP (F480) or YFP (F535) was collected. FRET was defined as the ratio of YFP fluorescence to YFP fluorescence (F535/F480). Details on optical filters and the correction of traces for photobleaching were described previously (13). To measure intracellular Ca2+ concentration, cells were illuminated at 490 nm, and emitted fluorescence was collected through the YFP channel between 520 and 550 nm. All fluorescent traces were recorded with a frequency of 5 Hz. Patch pipettes (5 to 10 megohm resistance) were pulled with borosilicate glass capillaries (GC150F, Harvard Apparatus) and a horizontal pipette puller (P-87, Sutter Instruments). All recordings were performed at ambient temperature. To generate concentration-response curves or voltage activation curves, data were plotted and fitted to the corresponding functions implemented in GraphPad Prism software (version 5, GraphPad Prism Software Inc.) or Origin 9.1 (Microcal). z values for M1-cam were calculated from the slope values of the sigmoidal Boltzmann equations (Fig. 1D) and were z = 0.76 (normal pipette solution) and z = 0.72 (GTPγS-containing pipette solution).

Chemicals and solutions

During all experiments, the cells were continuously superfused with extracellular buffer or agonist-containing buffer. The extracellular buffer was composed of 137 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 0.5 mM CaCl2, and 10 mM Hepes (pH 7.4). The intracellular pipette solution consisted of 100 mM K+-aspartate, 40 mM KCl, 5 mM NaCl, 7 mM MgCl2, 10 mM EGTA, 0.025 mM GTP, 5 mM Na+–adenosine triphosphate, and 20 mM Hepes (pH 7.2). The exchange of extracellular solutions was controlled by a solenoid valve–operated, pressurized perfusion system (ALA VC3-8SP, ALA Scientific Instruments) or by a gravity-driven, custom-made perfusion system (figs. S2, S5, and S6). For experiments with GTPγS, cells were loaded with 500 μM GTPγS through the patch pipette for 3 to 5 min to enable equilibration with the cytoplasm before the experimental protocol was applied. All agonists were purchased from Sigma or Tocris.

Molecular docking

Small molecules were docked into the x-ray crystal structure of wild-type M3 (27) or mutant receptors with DOCK3.6 software (3437). Anchor spheres to guide the placement of the molecules had been distributed on the basis of the molecular surface of the receptor and the pose of tiotropium in the x-ray structure of 4DAJ. The structure was prepared for docking such that ionizable side chains were charged, except for histidines, where protonation was modeled on the basis of the protein environment. Exploratory docking calculations for variants of active site residues were performed only once, whereas calculations for selected mutant and wild-type receptors were performed between three and five times with different random seed numbers to rule out the possibility that the ligand was trapped in a local minimum. The docking figures correspond to representative binding poses. A 180° flipping of the carboxyamide group of the mutated glutamine 6.52 did not affect the outcome of the calculations, with the exception of acetylcholine. In total, more than 100 docking calculations were performed to ascertain reliable binding modes.

Statistical analysis

Data are presented as individual observations or as means ± SEM of a given number of individual cells.


Fig. S1. Analysis of the voltage sensitivity of the M1 receptor when uncoupled from Gq proteins.

Fig. S2. Analysis of the voltage sensitivity of the M5 receptor.

Fig. S3. Variations of the calculated ligand binding poses.

Fig. S4. Analysis of the voltage sensitivity of the N6.52A and D3.32E mutant M3 receptors.

Fig. S5. Analysis of the voltage sensitivity of the choline-activated M3 receptor.

Fig. S6. Analysis of the voltage sensitivity of the N6.52Q mutant M1 receptor.

Fig. S7. The N6.52Q mutation in the M3 receptor reduces its affinity for ligands.

Table S1. Average distances between residues 6.52 of TM6 or 3.32 of TM3 and ligands in the wild-type and mutant M3 receptors.


Acknowledgments: We thank A. Galhoff and A.-L. Krett for excellent technical assistance and J. Holdich for the development of M1-cam. We also thank C. Taylor and N. Ma for carefully checking the statistics used in this study. Funding: This work was supported by grants from the German Research Foundation (DFG) (RI 1908/2-1) and FoRUM (F809-14) to A.R., KO 4095/1-1 to P.K., and SFB 593 TP A13 to M.B. P.K. and M.B. participate in COST (European Cooperation in Science and Technology) Action CM1207. Author contributions: A.R. designed and conducted cellular FRET experiments and analyzed the data; J.C.M. designed and performed molecular docking calculations and analyzed the data; M.M.-S. provided the M1-cam biosensor; P.K. designed docking calculations and analyzed the data; M.B. designed experiments and provided reagents and analytical tools; A.R. and M.B. wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Signaling

Navigate This Article