Identification of an Agonist-induced Conformational Change Occurring Adjacent to the Ligand-binding Pocket of the M 3 Muscarinic Acetylcholine Receptor *

TostudytheconformationalchangesthatconvertGprotein-cou-pled receptors (GPCRs) from their resting to their active state, we used the M 3 muscarinic acetylcholine receptor, a prototypical class A GPCR, as a model system. Specifically, we employed a recently developed insitu disulfidecross-linkingstrategythatallowsthefor-mation of disulfide bonds in Cys-substituted mutant M 3 muscarinic receptors present in their native membrane environment. At present, little is known about the conformational changes that GPCR ligands induce in the immediate vicinity of the ligand-binding pocket. To address this issue, we generated 11 Cys-substituted mutant M 3 muscarinic receptors and characterized these receptors intransfectedCOS-7cells.Allanalyzedmutantreceptorscontained an endogenous Cys residue (Cys-532 7.42 ) located within the exofacial segment of transmembrane domain (TM) VII, close to the ago-nist-binding site. In addition, all mutant receptors harbored a second Cys residue that was introduced into the exofacial segment of TM III, within the sequence Leu-142 3.27 –Asn-152 3.37 . Disulfide cross-linking studies showed that muscarinic agonists,

G protein-coupled receptors (GPCRs) 5 form one of the largest gene families found in nature. Various lines of evidence suggest that the human genome contains nearly 1000 genes coding for distinct GPCRs (1)(2)(3). GPCRs are cell-surface receptors that are activated by an extraor-dinarily diverse group of extracellular ligands (4 -7). Structurally, all GPCRs are composed of a bundle of seven transmembrane helices (TMs I-VII) that are connected by alternating intracellular and extracellular loops (see Fig. 1) (4 -7).
GPCR ligands recognize their target receptors by interacting with specific amino acids located on the extracellular receptor surface (4 -11). For example, the classic biogenic amine neurotransmitters bind to their target GPCRs within a cavity formed by the ring-like arrangement of the seven TMs (8 -11). In this case, the key residues contributing to ligand recognition are located primarily within the exofacial segments of TMs III and V-VII (8 -11).
Currently, bovine rhodopsin in its inactive state is the only GPCR for which high resolution structural information is available (12,13). Most GPCRs share a considerable degree of structural homology with rhodopsin and are therefore also referred to as rhodopsin-like or family A GPCRs. However, rhodopsin is unique among GPCRs in that its endogenous ligand, 11-cis-retinal, is covalently bound to the receptor protein and keeps rhodopsin in an inactive state by acting as an inverse agonist (14,15). It therefore remains unclear to what extent the activation mechanism of rhodopsin differs from that of GPCRs that are activated by diffusible ligands.
Following ligand binding, GPCRs undergo conformational changes that must be propagated from the ligand-binding domain to the intracellular receptor surface (5, 14 -17). The current view is that GPCR activation opens a cleft on the intracellular side of the receptor that enables the receptor to productively interact with heterotrimeric G proteins (5, 14 -17). Biophysical and biochemical studies carried out with bovine rhodopsin (14,15,17,18) and the ␤ 2 -adrenergic receptor (16, 19 -24) have revealed several activity-dependent conformational changes involving the intracellular receptor surface. The light-induced activation of rhodopsin is predicted to involve a reorientation of the cytoplasmic end of TM VI and changes in the relative disposition of TMs VI and III, along with smaller movements involving several other TMs (for recent reviews, see Refs. 14 and 15). Several studies suggest that similar conformational changes occur in GPCRs activated by diffusible ligands such as the ␤ 2 -adrenergic receptor (16, 19 -24).
Although much attention has been focused on identifying the conformational changes occurring on the intracellular receptor surface, little is known about the structural changes that GPCR ligands induce in the immediate vicinity of the ligand-binding domain (25). However, such information is essential for understanding how ligand binding to the extracellular receptor surface can lead to major structural changes on the cytoplasmic side of the receptor.
During the past few years, we have used the rat M 3 muscarinic receptor, a prototypical class A GPCR (9), as a model system to study activitydependent changes in GPCRs activated by diffusible ligands. Specifi-cally, we developed a novel in situ disulfide cross-linking strategy that allows the detection of disulfide bonds that can form between two Cys residues that are located adjacent to each other in the three-dimensional structure of the M 3 receptor protein (26). One major advantage of this approach is that Cys-substituted mutant M 3 muscarinic receptors can be characterized in their native membrane environment. In contrast, most studies examining activity-dependent changes in rhodopsin have been carried out with mutant proteins present in the solution state. However, as discussed by Hubbell et al. (15), the structural and dynamic properties of the solution state of rhodopsin may not be identical to those found in native disk membranes.
Recent in situ disulfide cross-linking studies showed that M 3 receptor activation is accompanied by pronounced conformational changes on the intracellular receptor surface (26,27). For example, we demonstrated that muscarinic agonists induce structural changes that increase the proximity of the cytoplasmic ends of TMs V and VI (26) and TMs I and VII (27). The observed cross-linking patterns also suggest that these conformational changes are accompanied by pronounced rotational movements of the cytoplasmic ends of TMs VI and VII (26,27).
The goal of this study was to identify agonist-induced conformational changes in the M 3 muscarinic receptor that occur in the immediate vicinity of the ligand-binding pocket. Classic muscarinic agonists are known to interact with their target receptors within a cleft enclosed by the ring-like arrangement of TMs I-VII, ϳ10 -15 Å from the membrane surface (9,11). The key amino acids involved in the binding of acetylcholine (ACh) and other classic muscarinic agonists are predicted to be located on the exofacial segments of TMs III and V-VII, including a TM III Asp residue that is conserved among all biogenic amine GPCRs (Asp-147 3.32 ; the superscript indicates the amino acid position according to the Ballesteros-Weinstein numbering system (28)) (see Fig. 1) and several muscarinic receptor-specific tyrosine residues (Tyr-148 3.33 , Tyr-506 6.51 , Tyr-529 7.39 , and Tyr-533 7.43 ) (see Fig. 1) (9,11).
In this study, we tested the hypothesis that muscarinic agonists, by simultaneously contacting residues present within TMs III and VII, can increase the proximity of the exofacial segments of TMs III and VII. To address this question, we took advantage of the fact that the modified version of the M 3 muscarinic receptor (referred to as M3Ј(3C)-Xa) that we developed as a template for Cys substitution mutagenesis contains only one single free Cys residue (Cys-532 7.42 ) within the TM receptor core (see Fig. 1). In agreement with a recently developed three-dimensional model of the M 3 muscarinic receptor (27), we demonstrated previously that Cys-532 is located adjacent to the ligand-binding pocket projecting into the center of the TM receptor core (29). Specifically, we examined whether Cys-532 is able to form activity-dependent crosslinks with Cys residues introduced into the exofacial segment of TM III (Leu-142 3.27 -Asn-152 3.37 ) (see Fig. 1).
We found that muscarinic agonists, but not antagonists, promoted the formation of a disulfide bond between Ser-151 and Cys-532 in the S151 3.36 C mutant receptor. Based on the predicted locations of Ser-151 and Cys-532 in the three-dimensional structure of the M 3 receptor protein, this finding strongly suggests that agonist activation of the M 3 muscarinic receptor increases the proximity of the exofacial segments of TMs III and VII. Because all class A GPCRs share a considerable degree of structural homology, a similar conformational change may occur in other members of this GPCR superfamily.

EXPERIMENTAL PROCEDURES
Materials-N-Ethylmaleimide, carbamylcholine chloride (carbachol), ACh bromide, atropine sulfate, and mammalian protease inhibitor mixture were purchased from Sigma. Acetyltriethylcholine (N-(Et 3 )-ACh) was generously provided by Dr.  Xa protease and digitonin were  from Roche Applied Science; and precast Novex Tris/glycine-polyacrylamide gels and SeeBlue Plus 2 prestained molecular mass standards  were from Invitrogen. Hybond TM ECL TM nitrocellulose membranes,  horseradish peroxidase-conjugated anti-rabbit IgG antibody, ECL TM  detection reagents, and Hyperfilm TM ECL TM chemiluminescence film were obtained from Amersham Biosciences. All other reagents used were of the highest grade commercially available.
Site-directed Mutagenesis-A pCD-based expression plasmid coding for a modified version of the rat M 3 muscarinic receptor, referred to as M3Ј(3C)-Xa (see Fig. 1), was used as a template for Cys substitution mutagenesis. As described previously (30), the M3Ј(3C)-Xa receptor contains an N-terminal hemagglutinin epitope tag and lacks all five potential N-terminal N-glycosylation sites and most endogenous Cys residues, except for Cys-140, Cys-220, and Cys-532 (see Fig. 1). Moreover, the central portion of the i3 loop (the third intracellular loop of GPCRs; Ala-274 -Lys-469) is replaced with two factor Xa cleavage sites (30). Cys residues were reintroduced into the M3Ј(3C)-Xa construct, one at a time, at Leu-142 3.27 -Asn-152 3.37 (see Fig. 1) using the QuikChange TM site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. The identity of all mutant constructs was verified by DNA sequencing.
Transient Expression of Cys-substituted Mutant M 3 Muscarinic Receptors in Mammalian Cells-All receptor constructs were transiently expressed in COS-7 cells. Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in a humidified 5% CO 2 incubator. Approximately 1 ϫ 10 6 cells were seeded into 100-mm dishes ϳ24 h prior to transfections. Cells were transfected with 4 g of receptor plasmid DNA/dish using the Lipofectamine Plus kit (Invitrogen) according to the manufacturer's instructions. To increase receptor expression levels, 1 M atropine was routinely added to the incubation medium for the last 24 h of culture as described previously by us (26,27,29).
Preparation of Membranes from Transfected COS-7 Cells-Cells were harvested ϳ48 h after transfections. To ensure complete removal of atropine that was included in the incubation medium during the last 24 h of culture, cells were washed twice (10 min each wash) with 10 ml of ice-cold phosphate-buffered saline (pH 7.4). After this washing step, 2 ml of ice-cold buffer A (25 mM sodium phosphate and 5 mM MgCl 2 (pH 7.4)) was added to each 100-mm dish, followed by a 15-min incubation at 4°C. Cells were then scraped off the plates and homogenized using a Polytron tissue homogenizer (setting 5, 20 s), followed by a 15-min centrifugation at 20,000 ϫ g at 4°C. The membrane pellets were then resuspended in buffer A (1 ml/100-mm dish), re-homogenized, frozen on dry ice, and stored at Ϫ70°C until used. Protein concentrations were measured using the Micro BCA protein assay reagent kit (Pierce) with bovine serum albumin as a standard.
Radioligand Binding Studies-Radioligand binding assays were carried out as described previously (26). In brief, membrane homogenates prepared from transfected COS-7 cells (ϳ10 -20 g of membrane protein/tube) were incubated in 1 ml of buffer A for 2 h at room temperature (22°C) in the presence of the muscarinic antagonist [ 3 H]NMS. In saturation binding assays, six different concentrations of [ 3 H]NMS (ranging from 20 to 3000 pM) were used. In competition binding assays, a fixed concentration of [ 3 H]NMS (500 pM) was employed in the presence of 10 different concentrations of the unlabeled competitor. Reac-tions were terminated by rapid filtration over Brandel GF/C filters, followed by three washes with ice-cold distilled water (ϳ4 ml/wash). Nonspecific binding was assessed as binding remaining in the presence of 1 M atropine. The amount of bound radioactivity was determined by liquid scintillation spectrometry. Binding data were analyzed using the nonlinear curve-fitting program PRISM 3.0 (GraphPad Software, Inc.).
Phosphatidylinositol (PI) Assays-To examine the ability of the different mutant receptors to productively interact with G proteins, we studied carbachol-induced increases in intracellular inositol monophosphate (IP 1 ) levels using transiently transfected COS-7 cells grown in 6-well plates (30). Cells were first labeled for 20 -24 h with myo-[ 3 H]inositol (3 Ci/ml) and then incubated in the presence of 10 mM LiCl for 1 h at 37°C with increasing concentrations of carbachol. The inositol monophosphate fraction was isolated and quantitated as described (30). Carbachol concentration-response curves were analyzed using the nonlinear curve-fitting program Prism 3.0.
Luciferase Assay-COS-7 cells were seeded in 24-well plates (ϳ50,000 cells/well) and cultured as described above. Approximately 24 h later, cells were cotransfected with 0.2 g of receptor DNA and 0.2 g of the luciferase reporter plasmid pAP-1-Luc (Stratagene) using the Lipofectamine Plus kit according to the manufacturer's instructions. The pAP-1-Luc plasmid contains the luciferase reporter gene driven by a basic promoter element (TATA box) joined to an activator protein-1 motif. About 48 h after transfections, cells were grown for 5 h in the presence of increasing concentrations of carbachol or N-(Et 3 )-ACh. Cells were then washed twice with phosphate-buffered saline and lysed with 0.5 ml of lysis buffer (25 mM glycylglycine, 15 mM MgSO 4 , 4 mM EGTA, 1 mM dithiothreitol, 1% Triton X-100, 15 mM KH 2 PO 4 , and 2 mM ATP (pH 7.8)). Cell lysates (100-l aliquots) were then transferred to 24-well flat bottom plates (Costar 3912) and mixed automatically with 125 l of reaction buffer (lysis buffer without Triton X-100) and 25 l of 0.8 mM luciferin in reaction buffer. Luminescence was then measured for 10 s using an Applied Biosystems TR717 microplate luminometer.
Cross-linking Studies-Receptor-containing membranes prepared from transfected COS-7 cells (see above) were thawed at room temperature and homogenized as described under "Preparation of Membranes from Transfected COS-7 Cells." To induce the formation of disulfide bonds, membranes prepared from one 100-mm dish (ϳ1 mg of protein present in a 1-ml volume) were incubated in microcentrifuge tubes with end-over-end rotation (30 rpm) for 10 min at room temperature with molecular iodine (20 M) in either the absence or presence of different concentrations of the muscarinic agonist carbachol or other muscarinic ligands. Reactions were terminated by the addition of EDTA and N-ethylmaleimide (10 mM each), followed by a 10-min incubation on ice.
To obtain membrane lysates (26), samples were then centrifuged at 8000 ϫ g for 10 min at 4°C, and the resulting membrane pellets were incubated with 250 l of 0.2% digitonin in phosphate-buffered saline (pH 7.4) for 20 min on ice to remove peripheral membrane proteins. Following another centrifugation step (8000 ϫ g for 10 min at 4°C), membrane pellets were incubated with 1.2% digitonin in 50 mM Tris-HCl (pH 8), 100 mM NaCl, and 1 mM CaCl 2 for 90 -120 min at 4°C with end-over-end rotation (30 rpm). After another centrifugation step (same conditions as described above), the supernatants (membrane lysates containing solubilized mutant M 3 muscarinic receptors) were transferred to fresh microcentrifuge tubes. Membrane lysates (ϳ15 g of protein) were then incubated with factor Xa protease (final concentration of 0.1 g/l) at room temperature for 16 -20 h (final volume of 50 l). The reactions were terminated by incubation for 30 min at room temperature with a mammalian protease inhibitor mixture (1:25 dilu-tion). Samples were then used directly for SDS-PAGE or stored at Ϫ70°C until used.
Western Blot Analysis-SDS-PAGE was performed essentially as described by Ward et al. (26). Samples were incubated for 30 min at 37°C with Laemmli loading buffer under nonreducing conditions and then loaded onto 10 -20% Tris/glycine-polyacrylamide gels, which were run at 125 V in the presence of 0.1% SDS. Western blot studies were carried out using the anti-M 3 antibody directed against the C-terminal 18 amino acids of the M 3 receptor protein (26,31). Receptor proteins were visualized using ECL detection reagents by autoradiography. The intensities of immunoreactive bands were quantitated by scanning densitometry using the program ImageQuant TL (Amersham Biosciences).
Molecular Modeling-We recently described a three-dimensional model of the TM core of the rat M 3 muscarinic receptor (27) that was built using homology modeling of the high resolution x-ray structure of the inactive state of bovine rhodopsin (12). All calculations were performed on a Silicon Graphics Octane workstation (300-MHz MIPS R12000 (IP30) processor) using the SYBYL 7.0 program (Tripos, Inc.). Overall, the M 3 muscarinic receptor model showed high structural similarity to that of the rhodopsin template (27).
ACh was constructed using the Sketch Molecule function of SYBYL 7.0 and subjected to a grid search, rotating all rotatable bonds by 60°i ncrements. Merck Molecular Force Field (32) and charge were applied using distance-dependent dielectric constants and the conjugate gradient method until the gradient reached 0.05 kcal/mol/Å. After clustering the low energy conformers resulting from the conformational search, the relative stabilities of various representative conformers from all groups were checked by semi-empirical molecular orbital calculations using the PM3 method in the MOPAC 6.0 package (33).
The lowest energy conformer of ACh was docked into the binding crevice of the rat M 3 muscarinic receptor using the previously established three-dimensional model of this receptor (27). Flexible docking was facilitated through the FlexiDock utility in the Biopolymer module of SYBYL 7.0. Taking into consideration the results of previous mutagenesis studies (9,11), ACh was prepositioned in the putative receptor binding site in a fashion similar to that described for the M 1 muscarinic receptor subtype (11). Flexible docking allowed for flexibility of all rotatable bonds in ACh and the side chains of the surrounding amino acids. To increase binding interactions, the torsion angles of the side chains that were located within 5 Å of the ligand (according to the results of FlexiDock) were adjusted manually. The structure of the AChreceptor complex was minimized using an AMBER force field with a fixed dielectric constant (4.0) until the conjugate gradient reached 0.05 kcal/mol/Å.

RESULTS
This study was designed to monitor agonist-induced conformational changes in the M 3 muscarinic receptor that occur close to the ligandbinding pocket. Toward this goal, we generated 11 Cys-substituted mutant M 3 muscarinic receptors and characterized these receptors in radioligand binding, functional, and disulfide cross-linking studies.

Generation of Cys-substituted Mutant M 3 Muscarinic Receptors-Single
Cys substitutions were introduced into a modified version of the rat M 3 muscarinic receptor (M3Ј(3C)-Xa) (30) that lacks most native Cys residues and contains two factor Xa cleavage sites within the i3 loop ( Fig.  1). Specifically, 11 consecutive residues in TM III (Leu-142-Asn-152) ( Fig. 1) were replaced, one at a time, with Cys residues. As a result, all generated mutant receptors contained two Cys residues within the TM receptor core: an endogenous Cys residue (Cys-532 7.42 ) located within the exofacial segment of TM VII and a second one newly introduced into TM III (Fig. 1).

Characterization of Mutant M 3 Muscarinic Receptors in Radioligand Binding and Functional Studies-All
Cys-substituted mutant M 3 muscarinic receptor constructs as well as the M3Ј(3C)-Xa "background" receptor were transiently expressed in COS-7 cells. Transfected cells were incubated with atropine (1 M) for the last 24 h of culture. Previous studies have shown that this atropine incubation step significantly increases the expression levels of Cys-substituted mutant M 3 muscarinic receptors (26,27,29).
To determine receptor densities, we carried out saturation binding studies with the muscarinic antagonist [ 3 H]NMS using membranes prepared from transiently transfected COS-7 cells. This analysis showed that the M3Ј(3C)-Xa receptor was expressed at a density of 2.85 Ϯ 0.14 pmol/mg of protein (B max ). The expression levels of most Cys-substituted mutant receptors differed from this value by Ͻ2.5-fold (TABLE  ONE). However, the L144C, S145C, and Y148C receptors showed an ϳ3-8-fold reduction in B max values (TABLE ONE).
To determine the affinity of the different mutant M 3 muscarinic receptors for the muscarinic agonist carbachol, we next carried out a series of [ 3 H]NMS/carbachol inhibition binding experiments. These studies showed that seven of the Cys-substituted mutant receptors displayed carbachol binding affinities (K i ) that differed from the carbachol affinity of the M3Ј(3C)-Xa construct (K i ϭ 35.0 Ϯ 15.8 M) by Ͻ2.2-fold (TABLE ONE). However, four mutant receptors (D147C, Y148C, A150C, and N152C) showed an ϳ3-9-fold reduction in carbachol binding affinities (TABLE ONE).
To examine whether the Cys-substituted mutant receptors were still able to couple to G proteins, we next studied their ability to mediate carbachol-induced increases in inositol monophosphate production (PI hydrolysis). Seven of the analyzed mutant receptors retained the ability to stimulate PI hydrolysis with high efficacy (E max , expressed as -fold increase in inositol monophosphate production above basal levels) (TABLE TWO). The V149C construct showed a pronounced reduction in G protein coupling efficacy (TABLE TWO). Strikingly, the S145C, D147C, and Y148C mutant receptors were completely devoid of functional activity (TABLE TWO).
In the PI assays, five of the analyzed mutant receptors (W143C, L144C, I146C, A150C, and N152C) showed clearly reduced (Ͼ9-fold) carbachol potencies (increased EC 50 values) compared with the M3Ј(3C)-Xa construct (EC 50 ϭ 20 Ϯ 11 nM) (TABLE TWO). The L144C construct showed by far the most pronounced decrease in carbachol potency, displaying an ϳ500-fold reduction in carbachol potency compared with the M3Ј(3C)-Xa receptor (TABLE TWO).  Eleven consecutive residues in TM III (Leu-142-Asn-152, underlined) were individually replaced with Cys residues. All Cys substitutions were introduced into a modified version of the rat M 3 muscarinic receptor referred to as M3Ј(3C)-Xa (30). Notably, the M3Ј(3C)-Xa construct contains only three remaining Cys residues, Cys-140, C-220, and Cys-532 (black boxes). All other native Cys residues are replaced with serine or alanine residues (white boxes). In addition, the central portion of the i3 loop (Ala-274 -Lys-469) is replaced with two factor Xa cleavage sites ((IEGR) 2 , underlined). To detect the different mutant M 3 muscarinic receptors via Western blotting, we used the rabbit anti-M 3 polyclonal antibody directed against the indicated C-terminal receptor sequence (31). Numbers refer to amino acid (aa) positions in the rat M 3 muscarinic receptor (59).
Disulfide Cross-linking Studies-As described above, transient expression of all analyzed Cys-substituted mutant M 3 muscarinic receptors resulted in the appearance of a significant number of [ 3 H]NMS-binding sites. Consistent with this observation, all mutant receptors could be easily detected in Western blot studies using membrane lysates prepared from transfected COS-7 cells and the anti-M 3 antibody polyclonal antibody directed against the C-terminal portion of the rat M 3 muscarinic receptor (data not shown).
To examine the potential spatial proximity of the Cys pairs present within the TM core of the 11 Cys-substituted mutant receptors, we studied their ability to form intramolecular disulfide bonds. All disulfide cross-linking studies were performed with mutant receptors present in their natural membrane environment (in situ) using membrane preparations obtained from transfected COS-7 cells (26,27,29). To promote the formation of disulfide bonds, receptor-containing membrane preparations were exposed to molecular iodine (20 M) during a 10-min incubation at room temperature. After this oxidation step, receptors were solubilized, digested to completion with factor Xa, and subjected to SDS-PAGE and Western blotting under reducing and nonreducing conditions. Under these conditions, as described in detail previously (26), the formation of an intramolecular disulfide bond will result in the appearance of an ϳ38-kDa receptor band under nonreducing conditions. This ϳ38-kDa immunoreactive species corresponds to properly folded cell-surface receptors, as demonstrated previously using surface biotinylation (26) and other biochemical techniques (29).
To examine whether the formation of disulfide bonds could be induced by receptor activation, cross-linking studies were carried out in the absence or presence of the muscarinic agonist carbachol. The outcome of these studies is summarized in Fig. 2. Like the M3Ј(3C)-Xa construct, nine of the Cys-substituted mutant M 3 muscarinic receptors analyzed did not yield a 38-kDa band under nonreducing conditions in either the absence or presence of carbachol (1 mM). However, two mutant receptors (A150C and S151C) gave a distinct cross-linking signal in both the absence and presence of the muscarinic agonist (Fig. 2).
Carbachol treatment had no significant effect on the intensity of the 38-kDa band in the case of the A150C construct. In contrast, carbachol simulation led to a pronounced increase in the intensity of the crosslinking signal in the case of the S151C mutant receptor (Fig. 2). When Western blot studies were carried out under reducing conditions, the 38-kDa bands were no longer observed (data not shown), indicating that these bands were not caused by incomplete digestion by factor Xa. The observed disulfide cross-linking signals required the exposure of receptor-containing membrane preparations to iodine (data not shown).
Agonist Dependence of the Disulfide Cross-linking Signal Observed with the S151C Receptor-To study the agonist dependence of the disulfide cross-linking signal displayed by the S151C mutant receptor in more detail, membranes prepared from cells expressing this mutant construct were oxidized in the presence of different carbachol concentrations (0.01-10 mM). As shown in Fig. 3A, carbachol treatment led to a concentration-dependent increase in the intensity of the 38-kDa cross-linking signal under nonreducing conditions. Fig. 3B summarizes the results of three independent carbachol cross-linking experiments carried out with the S151C mutant receptor based on the quantification of the intensity of the 38-kDa receptor species. The M3Ј(3C)-Xa construct was included in all experiments for control purposes.
The S151C Receptor Does Not Form Intermolecular Disulfide Bonds-The M 3 muscarinic receptor (34), like many other GPCRs (35,36), is known to form dimers or oligomers. We therefore wanted to exclude the possibility that agonist-promoted disulfide cross-linking between Cys-151 and Cys-532 occurred intermolecularly (between two or more receptor molecules) rather than intramolecularly. To address this issue, membranes prepared from S151C-expressing COS-7 cells were treated with iodine (20 M) in either the absence or presence of carbachol (1 mM). Samples were then processed for SDS-PAGE and Western blotting in the same fashion as described above under nonreducing conditions, except that the factor Xa cleavage step was omitted. Under these conditions, the formation of disulfide cross-links between receptor monomers is predicted to lead to the appearance of immunoreactive species of Ͼ75 kDa in size. However, Western blot studies showed that, like the M3Ј(3C)-Xa construct, the S151C mutant receptor did not yield significant amounts of high molecular mass receptor species in either the absence or presence of carbachol (Fig. 4). The Western blot studies

Functional properties of Cys-substituted mutant M 3 muscarinic receptors
The indicated Cys-substituted mutant M 3 muscarinic receptors were transiently expressed in COS-7 cells. All mutant receptors were derived from the M3Ј(3C)-Xa construct. To measure carbachol-induced increases in IP 1 levels, cells were preincubated for 20 -24 h with myo-[ 3 H]inositol (3 Ci/ml) and then stimulated in 6-well plates for 1 h at 37°C with increasing concentrations of carbachol (for details, see "Experimental Procedures"). Carbachol EC 50 and E max values were analyzed using the nonlinear curve-fitting program Prism 3.0. The data are given as means Ϯ S.E. of three independent experiments, each performed in duplicate.

Receptor
Carbachol-induced IP 1 production EC 50   revealed several immunoreactive bands: a major band corresponding to the receptor monomer and two considerably less intense bands migrating at ϳ40 and ϳ50 kDa, respectively. Although the molecular identity of these additional bands remains unclear at present, they may be due to cross-linking of the S151C mutant and M3Ј(3C)-Xa receptors to other membrane proteins. In any case, our data indicate that the formation of a disulfide bond between Cys-151 and Cys-532 in the S151C construct does not involve interactions between different receptor monomers.

Effect of Additional Muscarinic Ligands on Disulfide Bond Formation in the S151C Mutant
Receptor-To confirm that the carbachol-induced disulfide cross-link between Cys-151 and Cys-532 in the S151C receptor reflected a conformational change associated with receptor activation, we carried out additional cross-linking studies using several other muscarinic ligands (Fig. 5). Like the ACh derivative carbachol, ACh (1 mM) itself led to a pronounced increase in the intensity of the crosslinking signal when membranes prepared from S151C-expressing cells were subjected to oxidizing conditions (Fig. 6). No such effect was seen with the classic muscarinic antagonist atropine (1 mM) (Fig. 6).
A structural hallmark of all highly efficacious muscarinic agonists, including ACh and carbachol, is the presence of a compact ammonium head group (Fig. 5) (37,38). Replacement of the three N-methyl groups of ACh with larger alkyl substituents leads to partial muscarinic agonists or antagonists (37,38). In this study, we therefore also examined an ACh analog, N-(Et 3 )-ACh, in which all three methyl groups were replaced with ethyl moieties (Fig. 5). Radioligand binding studies showed that N-(Et 3 )-ACh retained the ability to bind to the S151C and M3Ј(3C)-Xa receptors, although with ϳ2-5-fold lower affinity than carbachol (TABLE THREE). To examine the functional properties of N-(Et 3 )-ACh, we used a luciferase reporter assay that is based on the ability of agonist-stimulated G q -coupled receptors to activate protein kinase C, which in turn triggers the activation of a basic promoter element (TATA box) containing an activator protein-1 motif, resulting in the expression of the luciferase reporter gene (39,40). We used this luciferase reporter assay rather than the PI assay to determine the functional activity of Et 3 -ACh because only a very small amount of N-(Et 3 )-ACh was available to us. (The luciferase assay could be carried out in a very small volume in a 24-well format.) As indicated in TABLE THREE, even the highest concentration of N-(Et 3 )-ACh used (1 mM) failed to simulate luciferase activity in COS-7 cells cotransfected with the S151C mutant or M3Ј(3C)-Xa receptor and the activator protein-1/luciferase reporter construct. However, under the same experimental conditions, carbachol stimulation of the S151C mutant or M3Ј(3C)-Xa receptor led to a pronounced increase in luciferase activity (TABLE THREE). These responses were characterized by carbachol EC 50 values that were in a similar range as the corresponding EC 50 values determined in PI assays (TABLE TWO). These data indicate that N-(Et 3 )-ACh can bind to the S151C and M3Ј(3C)-Xa receptors, but is unable to activate these receptors, characteristic of the behavior of a muscarinic antagonist. Consist-  Oxidative conditions have no effect on the electrophoretic mobility of the S151C mutant receptor. Membrane extracts prepared from COS-7 cells transfected with the M3Ј(3C)-Xa or S151C mutant receptor were treated with the oxidizing agent iodine (20 M) for 10 min at room temperature in either the absence or presence of carbachol (CCh; 1 mM). Samples were processed as described in the legend to Fig. 3, except that the factor Xa digestion step was omitted in this set of experiments (see "Experimental Procedures" for details). After solubilization of receptors with digitonin, samples (ϳ5 g of protein) were subjected to SDS-PAGE and Western blotting using the anti-M 3 antibody under nonreducing conditions. Note that incubation of the S151C mutant receptor (or M3Ј(3C)-Xa) with the oxidizing agent did not result in the formation of a significant amount of dimeric or oligomeric receptor species (see "Results" for more details). Protein molecular mass standards (M; in kilodaltons) are indicated to the right. ent with this concept, disulfide cross-linking studies showed that, like the classic muscarinic antagonist atropine, N-(Et 3 )-ACh failed to promote disulfide bond formation in the S151C mutant construct (Fig. 6).
Predicted Locations of Ala-150, Ser-151, and Cys-532 in the Threedimensional Structure of the M 3 Muscarinic Receptor-To facilitate the interpretation of disulfide cross-linking data, we recently built a threedimensional model of the TM core of the rat M 3 muscarinic receptor (27) using homology modeling based on the high resolution x-ray structure of the inactive state of bovine rhodopsin (12). As described in detail previously (27), the calculated model of the inactive state of the M 3 muscarinic receptor displayed high structural similarity to the rhodopsin template. Fig. 7 shows that Ala-150 3.35 , Ser-151 3.36 , and Cys-532 7.42 are located at about the same level within the TM receptor core (also see Fig. 1). Moreover, Fig. 7 illustrates that Ser-151 and Cys-532 directly face each other at the TM III/VII interface. Ala-150 is also located in the vicinity of Cys-532, but does not project toward Cys-532 directly.
We also used a molecular modeling strategy to dock ACh into the ligand-binding crevice of the rat M 3 muscarinic receptor (for details, see "Experimental Procedures"). Based on the results of previous mutagenesis studies (9,11), ACh was prepositioned in the binding site in a fashion similar to that described for the M 1 muscarinic receptor subtype (11). In agreement with the results of a recent NMR study examining the binding of two ACh analogs to the M 2 muscarinic receptor (41), ACh bound to the M 3 muscarinic receptor in the energetically preferred gauche conformation, displaying an O-C-2-C-1-N dihedral angle of about ϩ60°. Fig. 8 shows the location of Asp-147 3.32 and several Tyr residues (Tyr-148 3.33 , Tyr-506 6.51 , Tyr-529 7.39 , and Tyr-533 7.43 ) known to play key roles in ACh binding to the M 3 (9) and M 1 (11) muscarinic receptors. It also illustrates that Ala-150, Ser-151, and Cys-532 are located slightly below the plane defined by the bound ACh ligand. These observations, together with the results of the disulfide cross-linking studies, strongly suggest that agonist activation of the M 3 muscarinic receptor is associated with a conformational change that moves the exofacial portions of TMs III and VII closer to each other.

DISCUSSION
The ligand-dependent activation of GPCRs has been shown to lead to pronounced structural changes on the cytoplasmic surface of the receptor proteins (14 -17, 25). These conformational changes are thought to be essential for productive receptor/G protein coupling. At present, little is known about the molecular mechanisms by which ligand bind-ing to the extracellular receptor surface triggers the functionally critical conformational changes on the cytoplasmic side of the receptor. Thus, this study was undertaken to investigate whether diffusible ligands can induce conformational changes in the immediate vicinity of the ligandbinding pocket of a class A GPCR. To address this question, we employed a previously developed in situ disulfide cross-linking strategy (26,27,29) using the M 3 muscarinic receptor as a model system.
ACh and other classic muscarinic agonists are predicted to bind to the M 3 muscarinic receptor and other members of the muscarinic receptor family (M 1 -M 5 ) within a central binding crevice formed by the ring-like arrangement of the seven TMs (9,11). The amino acids that play a key role in ACh binding have been shown to be located within the exofacial segments of TMs III and V-VII (9,11). In general, other biogenic amine neurotransmitters are predicted to bind to their target GPCRs in a similar fashion (8,10).
The x-ray structure of bovine rhodopsin (12,13) and our newly generated M 3 muscarinic receptor model (27) suggest that the exofacial segments of TMs III and VII face each other in the three-dimensional structure of class A GPCRs. These receptor segments contain several residues that have been shown to be critically involved in ACh binding, including Asp-147 3.32 , Tyr-148 3.33 , Tyr-529 7.39 , and Tyr-533 7.43 (Figs. 1 and 8) (42,43). In this study, we therefore tested the hypothesis that agonist binding triggers changes in the relative orientation of the exofacial segments of TMs III and VII. Specifically, we generated 11 mutant M 3 muscarinic receptors, all of which contained a single endogenous Cys residue (Cys-532 7.42 ) within TM VII and a second Cys residue within the exofacial segment of TM III (Leu-142 3.27 -Asn-152 3.37 ). Biochemical (29) and molecular modeling (27) studies suggest that Cys-532 7.42 projects into the interior of the TM receptor core. We therefore considered Cys-532 7.42 a good "reporter" to detect agonist-induced changes occurring in the vicinity of the ligand-binding site.
Consistent with the known role of the exofacial segment of TM III in the binding of muscarinic ligands (9,11), several of the analyzed mutant receptors, including D147C and Y148C, exhibited reduced ligand binding affinities (TABLE ONE). Moreover, PI assays showed that three of the examined mutant receptors (S145C, D147C, and Y148C) lost the ability to interact with G proteins (TABLE TWO), indicating that these TM III residues play important roles in agonist-induced M 3 receptor activation. Similar results were obtained when the M 1 muscarinic receptor was subjected to alanine substitution mutagenesis (44).
To induce the formation of disulfide cross-links, the Cys-substituted mutant M 3 muscarinic receptors were exposed to oxidizing conditions in either the absence or presence of muscarinic agonists. In a previous study, we showed that molecular iodine, because of its relatively small size, is more efficient than the frequently used, more bulky oxidizing agent Cu(II)-phenanthroline in facilitating the formation of disulfide bonds between Cys residues present in the TM receptor core (29). In this study, we therefore used molecular iodine to promote the formation of disulfide bonds. Strikingly, disulfide cross-linking experiments led to the identification of a single mutant receptor (S151 3.36 C) that exhibited agonist-dependent disulfide bond formation. The S151C mutant receptor showed little cross-linking activity in the absence of ligands or in the presence of the muscarinic antagonist atropine (Figs. 2 and 6). In contrast, this construct displayed a pronounced, concentration-dependent cross-linking signal in the presence of the agonists carbachol and ACh (Figs. 2, 3, and  6). The S151C mutant receptor was able to bind muscarinic ligands with high affinity (TABLE ONE) and retained the ability to couple to G proteins with high efficiency (TABLE TWO). These observations clearly indicate that the cross-linking pattern observed with the S151C To exclude the possibility that agonist-promoted disulfide cross-linking between Cys-151 and Cys-532 involved the cross-linking of different receptor monomers, membranes prepared from S151C-expressing COS-7 cells were treated with iodine (20 M) in either the absence or presence of carbachol (1 mM) and then processed for SDS-PAGE and Western blotting under nonreducing conditions without cleaving the receptor with factor Xa. Under these conditions, like the M3Ј(3C)-Xa construct, the S151C mutant receptor did not yield significant amounts of high molecular mass receptor species in either the absence or presence of carbachol (Fig. 4), suggesting that the formation of a disulfide bond between Cys-151 and Cys-532 in the S151C construct most likely does not involve interactions between different receptor monomers. However, this analysis cannot completely rule out the possibility that disulfide-cross-linked receptor dimers do actually form, but that the C terminus of the receptor cannot be properly recognized by the anti-M 3 antibody in the structural context of the dimer.
Our three-dimensional model of the inactive form of the M 3 muscarinic receptor suggests that S151 3.36 C and Cys-532 7.42 face each other in the interior of the TM receptor bundle (Fig. 7). This model also predicts that the distance between the C-␣ atoms of Ser-151 and Cys-532 is ϳ8.2 Å. The distance between the C-␣ atoms of two Cys residues engaged in a disulfide bond usually ranges from ϳ3.8 to 6.8 Å (17, 45), providing a

Binding and functional activity of N-(Et 3 )-ACh and carbachol at the M3(3C)-Xa and S151C mutant receptors
Radioligand binding studies were carried out with membranes prepared from COS-7 cells transiently expressing the M3Ј(3C)-Xa or S151C mutant receptor. The binding affinities (K i ) for carbachol and N-(Et 3 )-ACh were determined in [ 3 H]NMS competition binding assays and were corrected for the Cheng-Prusoff shift (for details, see "Experimental Procedures"). For functional assays, COS-7 cells were cotransfected with receptor DNA (M3Ј(3C)-Xa or S151C) and an activator protein-1/luciferase reporter construct. Approximately 48 h after transfections, luciferase assays were carried out as described under "Experimental Procedures." Binding and functional data were analyzed using the nonlinear curve-fitting program Prism 3.0. Data are given as means Ϯ S.E. of three independent experiments, each performed in duplicate.   A three-dimensional model of the inactive state of the rat M 3 muscarinic receptor was built via homology modeling using the high resolution x-ray structure of bovine rhodopsin as a template (12,27). For the sake of clarity, only TMs II, III, VI, and VII are shown. Note that Ser-151 3.36 lies adjacent to Cys-532 7.42 at the TM III/VII interface. Ala-150 3.35 is also located in the proximity of Cys-532, but does not project toward Cys-532 directly. ACh (shown in orange) was docked into the ligand-binding crevice of the rat M 3 muscarinic receptor (lateral view) (for details, see "Experimental Procedures"). For the sake of clarity, only the exofacial segments of TMs II, III, VI, and VII are shown. Consistent with the outcome of a recent NMR study examining the binding of two ACh analogs to the M 2 muscarinic receptor (41), ACh bound to the M 3 muscarinic receptor in the energetically preferred gauche conformation, displaying an O-C-2-C-1-N dihedral angle of about ϩ60°. The highlighted aromatic amino acids (Tyr-148 3.33 , Tyr-506 6.51 , Tyr-529 7.39 , and Tyr-533 7.43 ) and Asp-147 3.32 are known to play key roles in ACh binding to the M 3 (9) and M 1 (11) muscarinic receptors. Note that Ala-150 3.35 , Ser-151 3.36 , and Cys-532 7.42 are located slightly below the ACh ligand.

Binding affinity (K
possible explanation as to why S151C and Cys-532 do not easily form a disulfide bond in the inactive state of the M 3 muscarinic receptor. The observed cross-linking pattern is therefore consistent with a model in which agonist binding to the M 3 muscarinic receptor increases the proximity of the exofacial segments of TMs III and VII, thus allowing the formation of a disulfide bridge between S151C and Cys-532. The cross-linking signal displayed by the S151C mutant receptor in the absence of ligands may be due to thermal motions of the polypeptide backbone of the transmembrane helical bundle (46). Consistent with this concept, disulfide cross-linking studies with Cys-substituted versions of a bacterial chemoreceptor of known structure indicate that disulfide bonds can form between two C-␣ atoms that are up to ϳ12 Å apart (46). Atropine treatment had no significant effect on the intensity of the basal cross-linking signal displayed by the S151C mutant receptor, suggesting that this signal is not due to constitutively active S151C mutant receptors.
In contrast to the S151C construct, the A150 3.35 C mutant receptor gave a weak cross-linking signal that remained essentially unchanged after the addition of the agonist carbachol (Fig. 2). Like Ser-151 3.36 , Ala-150 3.35 is also located in the vicinity of Cys-532 7.42 (estimated distance between the C-␣ atoms of Ala-150 and Cys-532 of 10.8 Å) (Fig. 7). However, in contrast to Ser-151, Ala-150 does not face Cys-532 directly, but projects toward TM II (Fig. 7), providing a possible explanation for the observed cross-linking pattern. Fig. 8 depicts a three-dimensional model of ACh docked onto the M 3 muscarinic receptor protein. This model illustrates that Ala-150 3.35 , Ser-151 3.36 , and Cys-532 7.42 are located slightly below the ACh ligand, explaining why ACh binding does not sterically block disulfide crosslinks between S151C (or A150C) and Cys-532.
As reviewed previously (9,11), the positively charged ammonium head group of ACh is predicted to lie adjacent to several aromatic amino acids located in TMs III, VI, and VII, including Tyr-148 3.33 , Tyr-506 6.51 , Tyr-529 7.39 , and Tyr-533 7.43 , and the negatively charged side chain of Asp-147 3.32 ( Figs. 1 and 8). In addition, a pair of TM V Thr residues (corresponding to Thr-231 5.39 and Thr-234 5.42 ) (Fig. 1) also contributes to high affinity binding of ACh to muscarinic receptors (42,43,47,48). (For the sake of clarity, TM V is not displayed in Fig. 8.) Site-directed mutagenesis studies have shown that these residues, which are conserved among all five members of the muscarinic receptor family (9,49), are critically involved in both ligand binding and receptor activation (9,11). Interestingly, a cluster of aromatic amino acids also plays an essential role in ACh binding to nicotinic ACh receptors (50). Protonated amines or quaternary amines bearing a formal positive charge, such as ACh and carbachol, can bind to aromatic side chains with high affinity via cation-interactions (51,52). Because the ammonium head group of ACh is surrounded by several aromatic residues (Tyr) in the muscarinic receptors (Fig. 8) (9, 11), it is likely that cation-interactions make a major contribution to ACh binding to this class of receptors.
As outlined above, our cross-linking data strongly suggest that ACh binding to muscarinic receptors results in a conformational change that moves the exofacial segments of TMs III and VII closer to each other. This finding is consistent with the concept that the aromatic "cage" that is a characteristic feature of the ACh-binding site is pulled closer to the ACh ammonium head group following ACh binding to the receptor protein (Fig. 8) (11). To further test this hypothesis, we examined an ACh analog, N-(Et 3 )-ACh, in which the three N-methyl groups of ACh were replaced with ethyl moieties, thus considerably increasing the size of the ammonium head group (Fig. 5). Classic structure-function activity studies have shown that substitution of the N-methyl groups of ACh with larger alkyl groups reduces the intrinsic efficacy of ACh and even-tually leads to muscarinic antagonists (37,38,53). Consistent with these findings, N-(Et 3 )-ACh retained the ability to bind to the M3Ј(3C)-Xa and S151C mutant receptors, but was no longer able to activate these receptors. Moreover, similar to the muscarinic antagonist atropine and in contrast to the agonists ACh and carbachol, N-(Et 3 )-ACh failed to promote the formation of a disulfide bridge between S151C 3.36 and Cys-532 7.42 (Fig. 6). Like N-(Et 3 )-ACh, atropine also contains a rather bulky ammonium group (Fig. 5). In contrast, high efficacy muscarinic agonists such as ACh and carbachol usually contain a compact quaternary head group (37,38,53). These data are consistent with the concept that muscarinic antagonists are unable to mimic the proposed AChinduced "tightening" of the aromatic cage around the ACh quaternary ammonium group, most likely due to steric hindrance caused by the increased size of their ammonium head groups.
Using a disulfide cross-linking strategy very similar to the one employed in this study, we previously demonstrated that agonist-mediated M 3 receptor activation leads to pronounced conformational changes on the intracellular receptor surface known to be involved in G protein recognition and activation (26,27). The observed pattern of disulfide cross-links, together with receptor modeling studies, strongly suggests that M 3 receptor activation increases the proximity of the cytoplasmic ends of TMs I and VII and TMs V and VI, associated with major rotational movements of the cytoplasmic portions of TMs VI (26) and VII (27). At present, the precise sequence of molecular events by which the agonist-induced conformational change observed in this study is propagated to the intracellular receptor surface remains unknown. However, because TM VII is engaged in multiple contacts with TM VI (12,13), it is possible that a movement of the exofacial segment of TM VII, as identified in this study, triggers conformational changes that are propagated to the endofacial segments of both TMs VI (26) and VII (27).
In agreement with our disulfide cross-linking data, site-directed mutagenesis studies with rhodopsin (54) and the C5a receptor (55) also suggest that activation of class A GPCRs is associated with changes in the relative orientation of TMs III and VII. In two related studies, metal ion-binding sites (i.e. His or Cys residues) were introduced into the exofacial segments of TMs III and VII at positions 3.32 and 7.39, respectively, of the ␤ 2 -adrenergic and tachykinin NK 1 receptors (56,57). Interestingly, Zn 2ϩ or Cu 2ϩ ions acted as partial agonists on the resulting mutant receptors, most likely by directly contacting the introduced Cys/His residues. Consistent with this concept, positions 3.32 and 7.39 face each other within the TM receptor core in the three-dimensional structure of the inactive state of rhodopsin (12). Although these findings are of considerable importance for a better understanding of GPCR function, it remains unclear whether the binding of metal ions to the Cys/His-substituted mutant receptors increases or decreases the proximity of the amino acids present at positions 3.32 and 7.39 (56,57).
In a previous study, Struthers et al. (58) used cysteine scanning mutagenesis and disulfide cross-linking in a split rhodopsin construct to investigate the proximity of Cys residues introduced at the o2 loop (the second extracellular loop of GPCRs)/TM V junction and the extracellular end of TM VI. Under mild oxidizing conditions, Cys residues introduced at positions 198 5.33 , 200 5.35 , and 204 5.39 (o2 loop/TM V junction) were able to form disulfide bonds with a Cys residue introduced at position 276 6.59 (top of TM VI) in the inactive state of rhodopsin. However, none of the three disulfide bonds interfered with 11-cis-retinal binding or light-induced rhodopsin activation (58). These data support the concept that the relative orientation of the extracellular ends of TMs V and VI is similar in the dark state and the light-activated state of rhodopsin (58). In contrast to the results of this study examining changes in the relative orientation of the exofacial segments of TMs III and VII, rhodopsin activation therefore does not seem to involve significant movements of the extracellular ends of TMs V and VI relative to each other.
To the best of our knowledge, this is the first study providing direct evidence for a conformational change occurring in the immediate vicinity of the binding pocket of a GPCR activated by a diffusible ligand. The extension of the approach described here to other TMs and residues that are buried more deeply in the TM receptor core should eventually provide a detailed view of the early conformational events triggering GPCR activation. Because all class A GPCRs share a high degree of structural homology, it is likely that the results that we obtained with the M 3 muscarinic receptor are also relevant for other members of this very large protein family.