Effects of N-ethylmaleimide on conformational equilibria in purified cardiac muscarinic receptors.

Muscarinic receptors purified from porcine atria and devoid of G protein underwent a 9-27-fold decrease in their apparent affinity for the antagonists quinuclidinyl benzilate, N-methylscopolamine, and scopolamine when treated with the thiol-selective reagent N-ethylmaleimide. Their apparent affinity for the agonists carbachol and oxotremorine-M was unchanged. Conversely, the rate of alkylation by N-ethylmaleimide, as monitored by the binding of [(3)H]quinuclidinyl benzilate, was decreased by antagonists while agonists were without effect. The receptor also underwent a time-dependent inactivation that was hastened by N-ethylmaleimide but slowed by quinuclidinyl benzilate and N-methylscopolamine. The destabilizing effect of N-ethylmaleimide was counteracted fully or nearly so at saturating concentrations of each antagonist and the agonist carbachol. Similar effects occurred with human M(2) receptors differentially tagged with the c-Myc and FLAG epitopes, coexpressed in Sf9 cells, and extracted in digitonin/cholate. The degree of coimmunoprecipitation was unchanged by N-ethylmaleimide, which therefore was without discernible effect on oligomeric size. The data are quantitatively consistent with a model in which the purified receptor from porcine atria interconverts spontaneously between two states (i.e. R R*). Antagonists favor the R state; agonists and N-ethylmaleimide favor the comparatively unstable R* state, which predominates after purification. Occupancy by a ligand stabilizes both states, and antagonists impede alkylation by favoring R over R*. Similarities with constitutively active receptors suggest that R and R* are akin to the inactive and active states, respectively. Purified M(2) receptors therefore appear to exist predominantly in their active state.

Muscarinic receptors purified from porcine atria and devoid of G protein underwent a 9 -27-fold decrease in their apparent affinity for the antagonists quinuclidinyl benzilate, N-methylscopolamine, and scopolamine when treated with the thiol-selective reagent N-ethylmaleimide. Their apparent affinity for the agonists carbachol and oxotremorine-M was unchanged. Conversely, the rate of alkylation by N-ethylmaleimide, as monitored by the binding of [ 3 H]quinuclidinyl benzilate, was decreased by antagonists while agonists were without effect. The receptor also underwent a time-dependent inactivation that was hastened by N-ethylmaleimide but slowed by quinuclidinyl benzilate and N-methylscopolamine. The destabilizing effect of N-ethylmaleimide was counteracted fully or nearly so at saturating concentrations of each antagonist and the agonist carbachol. Similar effects occurred with human M 2 receptors differentially tagged with the c-Myc and FLAG epitopes, coexpressed in Sf9 cells, and extracted in digitonin/ cholate. The degree of coimmunoprecipitation was unchanged by N-ethylmaleimide, which therefore was without discernible effect on oligomeric size. The data are quantitatively consistent with a model in which the purified receptor from porcine atria interconverts spontaneously between two states (i.e. R º R*). Antagonists favor the R state; agonists and N-ethylmaleimide favor the comparatively unstable R* state, which predominates after purification. Occupancy by a ligand stabilizes both states, and antagonists impede alkylation by favoring R over R*. Similarities with constitutively active receptors suggest that R and R* are akin to the inactive and active states, respectively. Purified M 2 receptors therefore appear to exist predominantly in their active state.
Sulfhydryl-specific reagents have been useful probes of the relationship between structure and function in muscarinic and other G protein-coupled receptors (e.g. Refs. [1][2][3][4][5]. In recent studies, such reagents have been tagged with environmentally sensitive fluorescent probes or spin labels and used to detect a conformational change linked to activation (1)(2)(3). That change is thought to involve the rotation or tilting of the sixth transmembrane helix (1,(3)(4)(5). The hydrophilic nature of some re-agents has been exploited to suggest that the conformational change leads to increased accessibility of the reactive residue (e.g. Refs. 4 and 5).
N-Ethylmaleimide is among the most widely used sulfhydrylspecific reagents. In the case of muscarinic receptors, as with many others, it has been shown to affect receptor-G protein coupling and to have a differential effect on the binding properties of agonists and antagonists. With receptors in native membranes, the Hill coefficients for the binding of muscarinic agonists are increased from characteristically low values to values near or indistinguishable from 1; moreover, the shift in potency brought about by guanyl nucleotides is reduced or abolished (e.g. Refs. 6 -9). The affinity of muscarinic antagonists generally is not affected by N-ethylmaleimide (7)(8)(9)(10)(11)(12), whereas the affinity of agonists can be either increased (8, 10 -14) or decreased (6,9).
It has been suggested that N-ethylmaleimide reacts preferentially with one of two spontaneously interconverting states of the receptor, thereby driving the equilibrium toward the reactive conformation (10 -12). Further support for the existence of multiple interconverting states, only one or comparatively few of which are functionally active, derives from the occurrence of constitutive activity as seen, for example, in mutants of the ␤ 2 adrenergic receptor (15) or the M 5 muscarinic receptor (16). In the latter case, it was demonstrated that the ligand-regulated activity of 13 constitutively active mutants, differing only at position 465 in helix number 6, can be accounted for by shifts in a single equilibrium between an inactive and an active state (16). Also, the notion of spontaneously interconverting states is consistent with the constitutive activity and related properties of M 1 -M 4 muscarinic receptors overexpressed in Chinese hamster ovary cells (17), and of M 5 muscarinic receptors coexpressed with comparatively large amounts of G q (18).
In the present study, N-ethylmaleimide has been used to probe for multiple states of cardiac muscarinic receptors purified from porcine atria and devoid of G protein. The effects of the reagent on the binding properties of the receptor, and the countervailing effects of muscarinic ligands on the action of N-ethylmaleimide, have been studied with the system under both thermodynamic and kinetic control. The data are quantitatively consistent with a model in which the receptor interconverts spontaneously between two states that are intrinsic to the receptor alone. N-Ethylmaleimide appears to favor the state that is of weaker affinity for antagonists and also undergoes a comparatively rapid inactivation, a pattern that is shared by constitutively active receptors (e.g. Refs. 4,19,and 20). MATERIALS 42.0 Ci/mmol; lot 3363333, 42.0 Ci/mmol) and from Amersham Biosciences (batch 44, 48 Ci/mmol). Both radioligands were supplied as a solution in ethanol. Both were devoid of synthetic precursors, as indicated by mass spectra provided by the manufacturers and, in the case of N-[ 3 H]methylscopolamine, obtained at the Molecular Medicine Research Centre of the University of Toronto. The absence of scopolamine from lot 3406009 of N-[ 3 H]methylscopolamine was confirmed by thin layer chromatography carried out by the manufacturer.
N-Ethylmaleimide was obtained from Sigma (ultra grade). Digitonin was obtained from Wako Bioproducts at a purity near 100% and at high purity from Roche Molecular Diagnostics and Calbiochem. The product from Wako generally was used to solubilize the receptor, and that from other sources was used to prepare and wash the columns used in various procedures. Sodium cholate was purchased from Sigma at a purity of 99% or better. Murine antibodies to the c-Myc epitope (9E10) conjugated to agarose were purchased from Santa Cruz Biotechnology, Inc. Murine antibodies to the FLAG epitope conjugated to horseradish peroxidase were from Sigma. Other materials were obtained from the sources identified previously (21).
Muscarinic Receptors-Purified M 2 receptor solubilized in digitonin/ cholate (0.1% digitonin, 0.02% sodium cholate) was prepared from porcine atria essentially as described previously (22). Receptor was extracted from the sarcolemmal fraction of the sucrose gradient according to the two-step procedure of Peterson and Schimerlik (23), except that the first step of the extraction was carried out with 0.36% digitonin and 0.07% sodium cholate. Subsequent passage through DEAE-Sepharose (Amersham Biosciences), 3-(2Ј-aminobenzhydryloxy)tropane-Sepharose, and hydroxyapatite (CHT-II, Bio-Rad) was carried out as described previously (21,22). The buffer used to elute the purified receptor from hydroxyapatite was exchanged for buffer D (20 mM KH 2 PO 4 , 20 mM NaCl, 1 mM Na 2 EDTA, 0.1 mM PMSF, 0.1% digitonin, and 0.02% sodium cholate, adjusted to pH 7.40 with KOH) (21) to obtain a stock solution that was divided into aliquots sufficient for one experiment and stored at Ϫ75°C.
The specific activity of the purified preparation was 5.4 mol/g of protein, as estimated from the maximal binding of [ 3 H]quinuclidinyl benzilate (QNB). 1 The corresponding purity was 28%, based on the molecular weight of 51,673 calculated from the primary sequence (24,25). The purified receptor migrated with the mobility of monomers, dimers, and larger oligomers during electrophoresis on polyacrylamide gels. Staining with silver nitrate and subsequent densitometry indicated that those bands represented 29% of all proteins visible on the gel in each of two such experiments. The identity of the bands was confirmed by means of controls that were blotted and then screened with receptor-specific antibodies. The purified receptor was devoid of the ␣ subunits of G o , G i1 , G i2 , G q/11 , and G s , as shown previously by immunoblotting with G protein-specific antibodies (22) and confirmed in the present investigation. Immunoblots with a nonspecific anti-caveolin antibody (BD Biosciences) demonstrated that the purified receptor also was devoid of caveolins 1, 2, and 3. 2 Human M 2 muscarinic receptors tagged with the c-Myc or FLAG epitope were co-expressed in Sf9 cells and extracted in digitonin/cholate (26). The final composition of the buffer was 15.7 mM KH 2 PO 4 , 15.7 mM NaCl, 0.78 mM Na 2 EDTA, 0.078 mM PMSF, 0.86% digitonin, and 0.17% sodium cholate, adjusted to pH 7.60 with KOH. The extract was stored at Ϫ75°C and used without further purification in subsequent assays. Complementary DNA coding for the wild-type and c-Myc-tagged human M 2 muscarinic receptor and cloned into a baculoviral vector was obtained from Biosignal, Inc. (Montreal). The construction of the baculovirus coding for FLAG-tagged human M 2 receptor has been described elsewhere (26).
Reaction with N-Ethylmaleimide-To prepare samples for use in the binding assays, an aliquot of purified receptor from porcine atria or the solubilized extract from Sf9 cells (28 -65 l) was thawed and diluted to about 450 l in buffer A (20 mM HEPES, 20 mM NaCl, 5 mM MgSO 4 , 1 mM Na 2 EDTA, 0.1 mM PMSF, 0.1% digitonin, and 0.02% sodium cholate, adjusted to pH 7.40 with NaOH) supplemented with N-ethylmaleimide. A fresh solution of N-ethylmaleimide was prepared for each reaction. Controls were prepared in the same manner except that Nethylmaleimide was omitted. The final concentration of N-ethylmaleimide in the reaction mixture was 10 mM except where stated otherwise; the concentration of receptor was 6.5-10 nM, as estimated from the binding of [ 3 H]quinuclidinyl benzilate to controls from which N-ethyl-maleimide was omitted. The reaction mixture was kept in an ice bath for 24 h, unless stated otherwise, and aliquots then were introduced directly into the binding assays. N-Ethylmaleimide was without effect on the binding of [ 3 H]quinuclidinyl benzilate and therefore was not removed in most experiments. All volumes were recorded, and the concentration of receptor was corrected as required in subsequent calculations.
When the reaction mixture was to contain N-ethylmaleimide together with either carbachol or N-methylscopolamine, the ligand was preincubated with the undiluted thawed receptor in an ice bath for 30 min. N-Ethylmaleimide then was added, dissolved in deionized water at 10 times the final concentration of 10 mM. The total volume of the reaction mixture was 20 -200 l. Controls were prepared in the same manner except for the omission of N-ethylmaleimide or the muscarinic ligand, as required. After the desired period of incubation in the ice bath, the volume of the reaction mixture was adjusted to 200 l with buffer A, and the sample was desalted on a column of Sephadex G-50 (fine) (0.8 ϫ 5.0 cm) previously equilibrated with buffer A. The eluate from the Sephadex column was used directly in the binding assays. To test for any effect of N-ethylmaleimide on the binding of [ 3 H]quinuclidinyl benzilate, samples prepared with the reagent were kept in an ice bath for 24 h, diluted with buffer A, and then desalted on Sephadex G-50 as described above.
Reconstitution-Native and alkylated receptors were reconstituted with G proteins according to a procedure adapted 3 from that described by Haga et al. (27). The G proteins were obtained as a mixture of isoforms purified from bovine brain (Calbiochem). Following incubation of the purified receptor for 24 h at 0°C with or without N-ethylmaleimide, as described above, the native or alkylated product (3.6 -7.3 pmol in 36 -80 l) was incubated with carbachol (10 mM) for about 15 min, also at 0°C. The receptor then was mixed (1:1) with a suspension of lipids (0.6 mg/ml L-␣-phosphatidylcholine, 0.6 mg/ml ␣-L-phosphatidyl-L-serine, and 60 g/ml cholesterol; Sigma) prepared in buffer B (20 mM HEPES, 160 mM NaCl, 6 mM MgCl 2 , 0.8 mM Na 2 EDTA, and 1 mM dithiothreitol, adjusted to pH 8.0 with KOH) supplemented with 0.18% sodium deoxycholate and 0.04% sodium cholate. The G proteins were added to the mixture (32-40 pmol of G␣ o , 8-16 pmol G␣ i-1 , 8-16 pmol of G␣ i-2 , Ͻ8 pmol of G␣ i-3 , and 64 pmol of G␤␥), which then was applied to a column of Sephadex G-50 (0.8 ϫ 5 cm) pre-equilibrated with buffer B. The column was eluted with buffer B, and the reconstituted receptors emerged in the void volume (450 l) at a concentration of 1.6 -4.9 nM. The vesicles were uniform in size with a mean diameter of 30 Ϯ 11 nm (n ϭ 6), as determined previously by electron microscopy. 3 Binding assays were performed on aliquots taken directly from the eluate of the Sephadex column.
Immunoprecipitation, Polyacrylamide Gel Electrophoresis, Western Blotting, and Immunodetection-The co-immunoprecipitation of c-Mycand FLAG-tagged M 2 receptors extracted from Sf9 membranes was carried out and monitored as described previously (26). Polyacrylamide gels that were to be silver-stained were fixed for 30 min in a solution of 50% methanol and 10% acetic acid. The solution was rinsed off, and the gel was warmed in a microwave oven (300 watt) for 1.5 min. The gel was incubated further with dithiothreitol (5 g/ml) for 5 min, rinsed, and stained for 30 min with a solution of silver nitrate (0.1%) (British Drug Houses). Bands were developed with 1.8% formaldehyde prepared in 3% sodium carbonate solution. As soon as the bands appeared, the development was terminated by incubating the gel with sodium citrate (2.3 M) for 10 min. Densitometric scanning was performed at a resolution of 300 dots per inch, and the intensity was determined using 1D Image Analysis software (Kodak Digital Science). Further details have been described elsewhere (26).
Binding Assays-Binding was measured essentially as described previously (21). Solutions of the radioligand and any unlabeled ligands were prepared in buffer A or buffer C (25 mM KH 2 PO 4 , 4 mM HEPES, 230 mM NaCl, 10 mM MgCl 2 , 0.8 mM Na 2 EDTA, and 0.1 mM PMSF, adjusted to pH 7.60 with KOH) for assays with solubilized or reconstituted receptor, respectively. An aliquot (5 l) then was added to the receptor (48 l) in polypropylene microcentrifuge tubes pretreated with trimethylchlorosilane. The reaction mixture was incubated at 30°C for the required period of time, and the bound radioligand was separated on columns of Sephadex G-50 (fine) (0.8 ϫ 6.5 cm). In assays with [ 3 H]quinuclidinyl benzilate, the time of incubation was 2 h unless indicated otherwise. Data obtained at graded concentrations of the radioligand (e.g. Fig. 1) were superimposable after incubation of the reaction mixture for either 2 or 6 h. Binding therefore was independent 1 The abbreviations used are: QNB, quinuclidinyl benzilate; GMP-P(NH)P, 5Ј-guanyly-5Ј-yl imidodiphosphate; NEM, N-ethylmaleimide; NMS, N-methylscopolamine; PMSF, phenylmethylsulfonyl fluoride. 2 (50 l) were removed at the desired times and applied to Sephadex G-50 as described above. Assays were performed in duplicate or triplicate, and the receptor was kept in an ice bath until mixed with the ligand and incubated at 30°C. Nonspecific binding was taken throughout as total binding in the presence of 1 mM unlabeled N-methylscopolamine. The value increased linearly with the concentration of unbound radioligand. Expressed as the fraction NS, as defined in Equations 1 and 4 below, the mean estimate of nonspecific binding from several representative experiments was 0.000080 Ϯ 0.000009 for [ 3 H]quinuclidinyl benzilate (n ϭ 19) and 0.000018 Ϯ 0.000002 for N-[ 3 H]methylscopolamine (n ϭ 16).
In assays with N-[ 3 H]methylscopolamine, ethanol that accompanied the radioligand accelerated the inactivation of receptors pretreated with N-ethylmaleimide. Most or all of the ethanol therefore was evaporated under argon before the radioligand was taken up in buffer. In such cases, the concentration of ethanol in the reaction mixture was less than 1% (v/v) at the highest concentration of N-[ 3 H]methylscopolamine and decreased proportionately at lower concentrations. Removal of ethanol had no effect on the binding of N-[ 3 H]methylscopolamine to untreated receptors or on the binding of [ 3 H]quinuclidinyl benzilate to either preparation. The radioligand therefore was used as supplied by the manufacturer in those assays.
Analysis of Data-All data were analyzed with total binding taken as the dependent variable (B obsd ) and with the total concentrations of all ligands taken as the independent variables. Subsequent manipulations did not alter the relationship between the data and the fitted curve. Except where stated otherwise, estimates of binding (B sp , B max ) and total receptor ([R] t ) denote the concentration in the binding assay (pM). The concentrations of ligands denote the total molar concentration in the binding assay.
Empirical analyses of the data were based on the Hill equation (Equations 1 or 2) and on Equation 3. The variables [P] t and [A] t represent the total concentrations of the radiolabeled probe and an unlabeled ligand, respectively. In Equation 1, B sp represents specific binding at the corresponding value of [P] t , and B max represents maximal specific binding; the parameter EC 50 denotes the concentration of unbound radioligand that yields half-maximal occupancy, and n H is the Hill coefficient. The parameter NS represents the fraction of unbound radioligand that appears as nonspecific binding (cf. Equation 14 in Ref. 28). Equation 1 was solved numerically (28). In Equations 2 and 3, the parameters B [A]ϭ0 and B [A]3ϱ represent the asymptotic levels of binding with respect to the concentration of unlabeled ligand; IC 50 is the concentration of unlabeled ligand that reduces specific binding by one-half. In Equation 3, F j represents that fraction of specific binding defined by the inhibitory potency IC 50(j) .
Semi-empirical and mechanistic descriptions of the data were based on Schemes I-III, in which there are one or more classes of mutually independent sites that bind either the radioligand (P) or an unlabeled ligand (A) in a mutually exclusive manner. It is assumed in Schemes I and II that the system is at thermodynamic equilibrium, whereas the formulation of Scheme III includes time as an explicit variable. In each case, the model was fitted to the data by means of Equation 4, in which the variables and parameters are as described above. An appreciable fraction of the radioligand bound to the receptor under at least some conditions in most experiments. Depletion therefore was accommodated in the calculation of B sp .
Scheme I comprises a mixture of distinct sites (R j , j ϭ 1, 2, . . . , n) wherein those of type j bind P and A with the equilibrium dissociation constants K Pj and K Aj , respectively, and constitute the fraction F j of all . Total specific binding of the probe was calculated according to Equation 5, and the required values of [PR j ] were obtained as described below. Values of EC 50 were calculated from the fitted estimates of K Lj (L ϵ P or A) and F j (n Ͼ 1), as described previously (21).
In Scheme II, each receptor of type j can exist in two spontaneously interconverting states designated R j and R j * . The ligands P and A bind to R j with equilibrium dissociation constants K Pj and K Aj (e.g.
[P][R j ]/ [PR j ] ϭ K Pj ), and the relative affinity of the ligand for the two states of the receptor is designated ␣ Pj and ␣ Aj , respectively (i.e. ␣ Pj ϭ K Pj */K Pj , . The parameter K Rj represents the relative concentrations of R j and R j * at equilibrium in the absence of ligand (i.e. [R j ]/[R j *] ϭ K Rj ). Sites of type j constitute the fraction F j of all sites, as in Scheme . The specific binding of the probe was calculated according to Equation 6, and the required values of [PR j ] and [PR j *] were obtained as described below.
In Scheme III, the receptor can exist in two spontaneously interconverting states (R 1 and R 2 ) that convert irreversibly to a third and fourth state (R 3 and R 4 , respectively). The equilibrium between R 1 and R 2 is analogous to that between R j and R j * for receptors of type j in Scheme . Analyses in terms of Schemes II and III required two classes of noninterconverting sites to accommodate both native and alkylated receptors. That heterogeneity is not shown explicitly in Scheme III, which depicts multiple states within a single class of sites, but it was accommodated in the model used in the calculations. The parameters k P(Ϫj) and k P(ϩj) in Scheme III denote the first-and second-order rate constants for the binding of a radioligand (P) to receptors in state j, and K Pj is the corresponding equilibrium dissociation constant (i.e. k P(Ϫj) /k P(ϩj) ϭ K Pj ). Similarly, the rate of interconversion between R 1 and R 2 is determined by k R(ϩ) and k R(Ϫ) (i.e. k R(Ϫ) /k R(ϩ) ϭ K R ), and that between PR 1 and PR 2 is determined by k PR(ϩ) and k PR(Ϫ) (i.e. k PR(Ϫ) /k PR(ϩ) ϭ K PR ). The rate of conversion of R j and PR j (j ϭ 1 or 2) to R jϩ2 and PR jϩ2 , respectively, is determined by k Rj and k PRj . The specific binding of the probe at any time t was calculated according to Equation  derived from the equations of state for P and A (e.g. Ϫ[P] t ϩ [P] ϩ B sp ϭ 0). Solutions were obtained according to the n-dimensional Newton-Raphson procedure (28).
The value of B sp in Equation 7 was calculated by numerical integration of the nine differential equations that define the system, which comprises eight states of the receptor and the unbound radioligand. Solutions were obtained according to the Bulirsch-Stoer method as implemented by Press et al. (29). The initial step size at each interval was taken as 1/100th of the half-time of the fastest process in the system or 1/100th of the interval to the next point, whichever was less. ] were taken as zero. The values of k P(ϩj) were provided by the fitting procedure, and those of k P(Ϫj) were computed as k P(ϩj) K Pj ; the values of K P3 and K P4 were selected arbitrarily, and the values of K P1 and K P2 were estimated in terms of Scheme II from data acquired at equilibrium (i.e. K Pj and ␣ Pj K Pj , respectively, in Scheme II). To ensure microscopic reversibility, the value of K PR was taken as K R K P2 /K P1 , and a fitted value was obtained for either k PR(ϩ) or k PR(Ϫ) . With data acquired at a given value of t and graded values of [P] t (i.e. Figs. 3B and 12), each point on the binding curve was treated as a separate time course beginning at t equals zero.
Statistical and Related Procedures-All parameters were estimated by nonlinear regression, and values at successive iterations of the fitting procedure were adjusted according to the algorithm of Marquardt (30). Equilibrium constants and related parameters were optimized throughout on a logarithmic scale (i.e. log K R , log K Pj , log ␣ Lj , etc.). Rate constants were optimized on a linear scale. Most analyses involved multiple sets of data, and data from different experiments were assigned separate values of [R] t and NS. Other details regarding the assignment of shared parameters are described in the legends to the figures and tables. The data from such analyses are presented with reference to a single fitted curve. To obtain the values plotted on the y axis, individual estimates of B obsd were adjusted for any differences in the parametric values from one set of data to another (i.e. Equation 6 in Ref. 31). Weighting of the data, tests for significance, and other statistical procedures were carried out as described previously (28,31,32).

Binding of Radiolabeled Antagonists and the Stability of Native and Alkylated
Receptors-Upon treatment with N-ethylmaleimide for 24 h at 0°C, M 2 receptors purified from porcine atria underwent a 20-fold reduction in their affinity for [ 3 H]quinuclidinyl benzilate and a decrease of 10 -30% in the corresponding capacity. The results from several experiments on two preparations of receptor are summarized in Table I, and representative data are illustrated in Fig. 1. There is no apparent relationship between the concomitant changes in EC 50 on the one hand and B max on the other. The changes in affinity appear to arise from alkylation per se, as described below, whereas the reduction in capacity may derive from the reduced stability of receptors treated with N-ethylmaleimide. Alkylation had similar effects on human M 2 receptors in digitoninsolubilized extracts from Sf9 cells (Table I). There was a 7-fold reduction in affinity as the apparent capacity decreased by almost 70%.
The affinity of the alkylated receptor for [ 3 H]quinuclidinyl benzilate was the same regardless of whether or not unreacted N-ethylmaleimide was removed on Sephadex G-50 immediately prior to the binding assay (Table I). It follows that no further alkylation occurred during the assay, assuming that the change in affinity reflects the progress of the reaction with the alkylating agent; also, N-ethylmaleimide appears to have no direct effect on the binding of the radioligand. Upon removal of the reagent, the capacity decreased on average from about 80 to 64% of that prior to alkylation. The further loss of sites may be because of incomplete recovery of the receptor from the desalting column. In most experiments, binding assays on the treated receptor were performed without removing unreacted N-ethylmaleimide. The effects of alkylation on affinity and capacity were essentially the same when the time of incubation at 0°C was extended from 24 h to 4 days. The reaction therefore was complete after 24 h, which was the interval used in most of the experiments described below. Finally, the effects of N-ethylmaleimide were the same when the concentration was reduced from 10 to 1 mM over 24 h at 0°C (Table I).
The time-dependent binding of [ 3 H]quinuclidinyl benzilate to native and alkylated receptors from porcine atria is illustrated in Fig. 2. Binding to the untreated receptor was faster at higher concentrations of the radioligand, becoming virtually instantaneous at saturating concentrations. Binding to the alkylated receptor was instantaneous at all concentrations of the radioligand used in the assays. Under all conditions, binding remained stable for up to 4 h after the maximal level was attained.
Alkylation reduced the apparent affinity of purified receptors for N-[ 3 H]methylscopolamine (Table II, Fig. 3), 4 consistent with the effect on [ 3 H]quinuclidinyl benzilate, but the magnitude of the change increased with the time of incubation with the radioligand. Incubation for 15 min caused a 10-fold increase in EC 50 , from 2.7 nM with the native receptor to 28 nM after alkylation; incubation for 3 h caused a 30-fold increase, from 2.7 to 85 nM. The relative capacity for N-[ 3 H]methylscopolamine and [ 3 H]quinuclidinyl benzilate was about 95% with the native receptor and decreased appreciably, to about 77%, after treatment with N-ethylmaleimide (Table II,  In assays with N-[ 3 H]methylscopolamine, the time-dependent nature of the effect on EC 50 suggests that the alkylated receptor was unstable under those conditions. To confirm that the increase in EC 50 was not wholly a consequence of instability, but derived at least in part from a decrease in affinity per se, the binding of N-[ 3 H]methylscopolamine was measured after equilibration for 24 h at 0°C. In three such experiments, with native and treated receptor taken in parallel, the mean value of log EC 50 increased upon alkylation from Ϫ9.02 Ϯ 0.07 to Ϫ8.14 Ϯ 0.42. The corresponding estimates of capacity are 551 Ϯ 97 and 481 Ϯ 100 pM for the native and alkylated 4 The data shown in Figs. 3 and 7 were analyzed in terms of Scheme I, and the parametric values are listed in Tables II and III, respectively. The data in Fig. 3B also were combined with those in Fig. 4 and analyzed in terms of Scheme III, as described under "Discussion" and in the legend to Fig. 3. Similarly, the data represented in Fig. 7 also were combined with those in Fig. 9A and analyzed in terms of Scheme II, as described under "Discussion" and in the legends to Figs. 7 and 9. The use of Scheme I is essentially empirical, because the model offers no explanation for the interplay between N-ethylmaleimide on the one hand and muscarinic ligands on the other. In contrast, Schemes II and III describe the system in mechanistically explicit terms. SCHEME 3

N-Ethylmaleimide and Purified Muscarinic Receptors
receptor, respectively. N-Ethylmaleimide therefore reduced the affinity for both [ 3 H]quinuclidinyl benzilate and N-[ 3 H]methylscopolamine by a similar amount, the relative instability of the receptor in the presence of the latter notwithstanding. N-[ 3 H]Methylscopolamine bound rapidly to native and alkylated receptors at all concentrations of the radioligand, but the initial burst was followed by a loss of binding upon continued incubation (Fig. 4). The latter effect was irreversible, as de-scribed below, and such a loss is in contrast to the stability observed in the binding of [ 3 H]quinuclidinyl benzilate (Fig. 2). The decay was more rapid at lower concentrations of N-[ 3 H]methylscopolamine, which therefore appears to protect the receptor from inactivation. The decay also was more rapid after alkylation, as illustrated in Fig. 4 where the concentrations of the radioligand were selected to achieve comparable levels of occupancy in the two preparations.
To examine their intrinsic stability, native and alkylated receptors were preincubated at 30°C for different periods of time prior to the addition of a saturating concentration of [ 3 H]quinuclidinyl benzilate. With the alkylated receptor, approximately one-half of the initial capacity was lost within 5 min, and virtually all activity was lost within 1 h (Fig. 5). With the native receptor, the half-time of inactivation varied from at least 100 min to much longer; in some experiments, binding was essentially stable for at least 3 h. When the alkylated receptor was preincubated in the presence of 3.9 nM [ 3 H]quinuclidinyl benzilate, binding declined rapidly but leveled off at a value that was consistent with the degree of labeling attained during preincubation (Fig. 5). Bound [ 3 H]quinuclidinyl benzilate therefore protected the receptor from the inactivation that otherwise occurred in its absence, in accord with the stability evident in Fig. 2. There was a similar stabilization of the native receptor when [ 3 H]quinuclidinyl benzilate was present during preincubation.
The inactivation that occurred at 30°C was irreversible, as indicated by the failure to regain binding when the temperature was returned to 0°C. Alkylated receptor was preincubated for 20 min at 30°C and then transferred to an ice bath for a further 24 h, all in the absence of ligand. At each step, binding was measured at graded concentrations of [ 3 H]quinuclidinyl benzilate. The capacity decreased from 477 pM initially to 42.4 pM after preincubation; 24 h later, it was essentially unchanged at 31.3 pM. Lowering the temperature therefore prevented further inactivation but failed to retrieve those sites already lost.
The instability caused by N-ethylmaleimide was prevented by the agonist carbachol, which had a protective effect similar to that of N-methylscopolamine or quinuclidinyl benzilate. When the alkylated receptor was preincubated at 30°C for 15 min, the capacity for [ 3 H]quinuclidinyl benzilate was reduced to 19% of that in a control preincubated on ice (Fig. 6, cf. Fig. 5). In contrast, the capacity was virtually identical to that of the control (p ϭ 0.25) when 10 mM carbachol was included during the preincubation at 30°C and then removed prior to the binding assays.  (21). Individual estimates of log EC 50 and n H were averaged to obtain the mean (Ϯ S.E.) listed in the table. Individual estimates of B max were adjusted for dilutions prior to the binding assay to obtain an imputed value for the stock solution. The mean is shown relative to that for unreacted receptor from the same batch taken as 100. The absolute values of B max (nM) for the two batches of purified receptor are as follows (Ϯ S.E.): 1, 12.7 Ϯ 0.52 (6) Alkylated receptor (‚, ‫,ء‬ ϩ) and a control from which NEM was omitted (E, छ, Ⅺ) were prepared as described under "Materials and Methods." Binding to the two preparations was measured in parallel at graded concentrations of the radioligand, either alone (upper curves) or together with 1 mM NMS (baseline). Different symbols denote data from three different experiments (E, ‚; छ, ‫;ء‬ Ⅺ, ϩ) carried out on samples from two batches of purified receptor. The lines represent the best fit of Equations 4 and 5 (Scheme I) to all of the data taken together. One class of sites was sufficient throughout. With the native receptor, there was no appreciable increase in the sum of squares with 1 rather than 3 values of K P1 (p ϭ 0.25) (log K P1 ϭ Ϫ9. 43  Binding of Unlabeled Ligands-Native and alkylated receptors were examined for the inhibitory effect of five unlabeled ligands, three antagonists and two agonists, on the binding of [ 3 H]quinuclidinyl benzilate (Fig. 7). To define all parameters in subsequent analyses and to test for internal consistency, binding also was measured at graded concentrations of [ 3 H]quinuclidinyl benzilate and N-[ 3 H]methylscopolamine (Fig. 7, A and  D). In assays with the alkylated receptor and N-[ 3 H]methylscopolamine, the time of incubation was 15 min to avoid or at least to minimize inactivation.
The data represented in all panels of Fig. 7 were pooled and analyzed in terms of Scheme I (n ϭ 2), and the parametric values are listed in Table III. 4 Although the fit was better with two classes of sites rather than one, the difference is not readily discernible in the fitted curves. Like [ 3 H]quinuclidinyl benzilate and N-[ 3 H]methylscopolamine, both scopolamine and unlabeled N-methylscopolamine bound more weakly to alkylated receptors than to the native preparation (i.e. ⌬log EC 50 Յ Ϫ1.3, Table III). A substantial reduction in affinity therefore appears to be a common effect of N-ethylmaleimide on the binding of antagonists. In contrast, there was little if any change in the affinities of the agonists carbachol and oxotremorine-M (Fig. 7, C and F; Table III).
The decreased affinity of unlabeled antagonists indicates that the corresponding decrease in the affinity of [ 3 H]quinuclidinyl benzilate and N-[ 3 H]methylscopolamine is not a kinetic artifact of destabilization. Because of the protection afforded by muscarinic ligands, one of which is always present in assays at near saturating concentrations of [ 3 H]quinuclidinyl benzilate, the inhibitory potency is expected to be unaffected by the instability of the vacant receptor after alkylation. Furthermore, the affinity of quinuclidinyl benzilate was the same when estimated from binding at graded concentrations of [ 3 H]quinuclidinyl benzilate (Fig. 7D) and when inferred from the inhibitory effect of the unlabeled analogue ( Fig. 7E) (Table III). Although there was some discrepancy between labeled and unlabeled N-methylscopolamine, N-ethylmaleimide reduced the affinities of both analogues by a similar amount. These observations support the conclusions that emerge from Scheme I and the parametric values listed in Table III: namely, that the effect of N-ethylmaleimide on the binding of antagonists derives from a change in affinity per se, whereas the affinity of agonists is essentially unchanged.
The binding of oxotremorine-M was sensitive to GMP-P(NH)P when purified M 2 receptor was reconstituted with G o/i in phospholipid vesicles (Fig. 8). Reconstitution alone led to a more pronounced heterogeneity, and the Hill coefficient for the native receptor decreased from 0.86 Ϯ 0.05 in solution (n ϭ 3, Fig. 7C) to 0.58 Ϯ 0.04 in vesicles (n ϭ 2, Fig. 8). With the alkylated receptor, n H decreased from 0.88 Ϯ 0.02 (n ϭ 3) in solution to 0.59 Ϯ 0.02 (n ϭ 2) in vesicles. There was no effect of G o/i on the binding of oxotremorine-M when receptors and G proteins were mixed in solution. 3 GMP-P(NH)P increased the Hill coefficient to 0.77 Ϯ 0.15 and 0.68 Ϯ 0.08 with native and alkylated receptors, respectively. The data required two classes of sites when described empirically in terms of Equation 3, and the fitted curves are illustrated in Fig. 8. The overall effect of GMP-P(NH)P was estimated as the area between the curves   Fig. 3B taken together. Single values of k P(ϩj) , k P(Ϫj) , k R1 , k PR1 , and k PR2 were common to all data acquired with native and alkylated receptors, which therefore differed only in the values of k R2 and the rate constants that define K R and K PR . To ensure consistency with the results obtained at equilibrium, the values of log K P1 and log K P2 were fixed at Ϫ9.87 and Ϫ7.69, respectively (i.e. log K L and log (␣ L K L ) for [ 3 H]NMS in terms of Scheme II (Table IV)); similarly, the value of log K Rj was fixed at Ϫ1.28 and Ϫ3.28 for native and alkylated receptors, respectively (Table V). The values of log K P3 and log K P4 were fixed arbitrarily at Ϫ3 to preclude appreciable binding at the highest concentration of [ 3 H]NMS. The fit is independent of the values of the rate constants for the interconversion between R 1 and R 2 , or between PR 1 and PR 2 , provided that each step is rapid on the time scale of the binding assay. The fitted estimates of the rate constants for inactivation are as follows (min Ϫ1 ): k R1 (native and alkylated receptor) ϭ 0.00008 Ϯ 0.0025, k PR1 (native and alkylated receptor) ϭ 0.00014 Ϯ 0.00045, k R2 (native receptor) ϭ 0.0024 Ϯ 0.0010, k R2 (alkylated receptor) ϭ 0.013 Ϯ 0.001, k PR2 (native and alkylated receptor) ϭ 0.00053 Ϯ 0.00033. The fitted estimates of other parameters are as follows (M Ϫ1 min Ϫ1 ): k P(ϩ1) ϭ 5.0 ϫ 10 8 , k P(ϩ2) ϭ 5.6 ϫ 10 7 , k P(ϩ3) ϭ 2.9 ϫ 10 3 , and k P(ϩ4) ϭ 3.0 ϫ As illustrated in Fig. 9A, N-methylscopolamine slowed the rightward shift induced by N-ethylmaleimide in the binding curve for [ 3 H]quinuclidinyl benzilate. When the data were analyzed collectively in terms of Scheme I (Equation 5, n ϭ 2), the shift can be described as a time-dependent interconversion of sites from a state of higher affinity (log K P1 ϭ Ϫ9.31 Ϯ 0.02) to a state of lower affinity (log K P2 ϭ Ϫ8.29 Ϯ 0.03). Native receptors were wholly in the state of higher affinity. After treatment for 30 min in the absence of antagonist, the low affinity sites represented about 60% of the total binding (F 2 ϭ 0.62 Ϯ 0.02); after 30 min in the presence of N-methylscopolamine, they represented less than 40% of the total binding (F 2 ϭ 0.36 Ϯ 0.03). After more prolonged treatment in the absence and presence of N-methylscopolamine, the low affinity sites represented more than 80% (4 h, F 2 ϭ 0.84 Ϯ 0.02; 24 h, F 2 ϭ 0.83 Ϯ 0.02) and about 60% (4 h, F 2 ϭ 0.61 Ϯ 0.02; 24 h, F 2 ϭ 0.59 Ϯ 0.02) of the total binding, respectively.
The rate of the reaction with N-ethylmaleimide was unaffected by carbachol. As illustrated in Fig. 9B, the binding profiles obtained after alkylation for 30 min with and without the agonist are virtually superimposable. In terms of Scheme I, about 70% of the sites were in the state of lower affinity for [ 3 H]quinuclidinyl benzilate in either case (no carbachol, F 2 ϭ 0.67 Ϯ 0.01; with carbachol, F 2 ϭ 0.71 Ϯ 0.01).
Oligomeric Status of M 2 Receptor Extracted from Sf9 Cells-G protein-coupled receptors are known to occur as oligomers (22,26,33). The relationship between oligomeric status and binding therefore was examined in extracts from Sf9 cells coexpressing the c-Myc-and FLAG-tagged M 2 receptors. Because the FLAG epitope coimmunoprecipitated with the c-Myc epitope, as identified on Western blots (Fig. 10, A and B), at least some of the receptors were present as dimers or larger oligomers in vivo. No signal is obtained if membranes from cells infected separately with the two baculoviruses are mixed prior to solubilization (26). The coprecipitated receptor migrated primarily as two immunoreactive bands with molecular masses indicative of monomers (59.5 Ϯ 1.6 kDa, n ϭ 6) and dimers (93.4 Ϯ 2.6 kDa, n ϭ 6), based on the calculated value of 51,673 Da for a monomer (24,25). Some blots also contained a minor band exhibiting a molecular mass of 165 Ϯ 19 kDa (n ϭ 3), perhaps indicating a trimer.
Tagged receptors extracted from Sf9 cells were stable at 30°C, because preincubation for 20 min affected neither the capacity for [ 3 H]quinuclidinyl benzilate nor the degree of coimmunoprecipitation (Fig. 10). N-Ethylmaleimide was without effect on the degree of coimmunoprecipitation, which was unchanged after preincubation at 30°C in the absence of ligand or in the presence of either quinuclidinyl benzilate or N-methylscopolamine (Fig. 10). The alkylated receptor was functionally  unstable, however, in that about 60% of the original sites were lost before or during the binding assay (Table I, Fig. 10C). The overall loss increased to 86% after preincubation for 20 min and to 97% after preincubation for 1 h (Fig. 10C).
Alkylation by N-ethylmaleimide was similar in its effect on the binding properties of M 2 receptors from Sf9 cells and porcine atria. Both preparations underwent a reduction in their affinity for [ 3 H]quinuclidinyl benzilate (Table I) and an increase in the rate of inactivation at 30°C (cf. Figs. 5 and 10C), although the latter effect was much greater with the recombinant receptor. Because N-ethylmaleimide was without effect on the degree of coimmunoprecipitation from Sf9 extracts, the loss of function apparently was not accompanied by any change in the oligomeric status of the receptor.

Effects of N-Ethylmaleimide on the M 2 Muscarinic
Receptor-Muscarinic ligands and the sulfydryl-specific reagent N-ethylmaleimide exhibit a pattern of complementary effects at M 2 receptors purified from porcine atria. Alkylation caused a 9-27-fold reduction in the affinities of three muscarinic antagonists, as estimated in terms of Scheme I, whereas the affinities of two agonists were unchanged. Conversely, the reaction with N-ethylmaleimide was slowed by the antagonist N-methylscopolamine but not by the agonist carbachol. The receptor also underwent a time-dependent, irreversible loss of capacity that was faster after alkylation and slower in the presence of quinuclidinyl benzilate or N-methylscopolamine. The destabilizing effect of N-ethylmaleimide was counteracted fully or nearly so at saturating concentrations of either antagonist or the agonist carbachol. The protective effect of  Table  II (native receptor, 0.95; alkylated receptor, 0.77). There were fewer constraints with Scheme I (n ϭ 2) than with Scheme II, and the sum of squares is 15% less for the data represented in Fig. 7, but the fitted curves from Scheme I are almost superimposable with those shown in the figure.
( Fig. 3B), as described below, the major change occurred prior to the binding assays. Moreover, the decrease in affinity was retained under conditions when the alkylated receptor was comparatively stable. [ 3 H]Quinuclidinyl benzilate afforded more protection than did N-[ 3 H]methylscopolamine (Figs. 2   and 4), yet the change in affinity was similar or the same for both radioligands (Table III). Also, there was a comparable change in the affinities of three unlabeled antagonists, as inferred from their inhibitory effect under conditions that are expected to preclude inactivation. In binding assays conducted overnight at 0°C, treatment with N-ethylmaleimide was found to reduce the affinity but not the capacity of purified receptors for N-[ 3 H]methylscopolamine. The independent nature of the changes in affinity and capacity is illustrated further by a comparison of receptors from Sf9 cells and porcine atria. Whereas the destabilizing effect of alkylation was markedly greater with the Sf9 extract, the decrease in affinity for [ 3 H]quinuclidinyl benzilate was similar in the two preparations.
It has been reported that N-ethylmaleimide can react with G proteins, thereby precluding their interaction with the receptor (34) and inhibiting GTPase activity (35). Because the purified receptor was devoid of ␣ o , ␣ i1 , ␣ i2 , ␣ q/11 , and ␣ s , the effects of alkylation found in the present investigation were independent of G proteins; rather, the observed changes in binding and stability apparently derived from an effect intrinsic to the receptor alone. When the purified receptor was reconstituted with G o/i in phospholipid vesicles, the binding of agonists was affected by N-ethylmaleimide in the manner that is characteristic of muscarinic receptors in native membranes (e.g. Refs. 6, 8, and 9). In particular, the reagent decreased the magnitude of the effect of GMP-P(NH)P. The purified receptor therefore retained its native functionality with respect to the interaction with G o/i and the allosteric interaction between the agonist and GMP-P(NH)P; also, the related effects of N-ethylmaleimide are largely independent of whether alkylation occurs in the native membrane or after purification.
N-Ethylmaleimide did not affect the nature or quantity of oligomers formed by differently tagged human M 2 receptors, at least as monitored by the coimmunoprecipitation of c-Myc and FLAG epitopes from extracts of coinfected Sf9 cells. Cloned receptors and receptors from porcine atria underwent qualitatively similar changes in affinity and stability, with the two preparations differing only in the magnitude of the change. It therefore appears that neither the instability nor the decreased affinity for [ 3 H]quinuclidinyl benzilate arose from any effect of alkylation on the oligomeric status of the receptor.
Evaluation in Terms of Scheme II-Whereas the changes in affinity and stability appear to be mutually independent, the reciprocal nature of the various effects argues for a common cause. Many studies have shown that muscarinic and other G  Fig. 7 were analyzed simultancously in terms of Scheme I (Equations 4 and 5, n ϭ 2) to obtain the parametric values for all ligands (radiolabeled, L ϵ P; unlabeled, L ϵ A). All experiments were performed at least three times except for the inhibition of [ 3 H]QNB by unlabeled QNB in panel E, which was done once. Assays at two concentrations of [ 3 H]QNB generally were performed in parallel (panels B and E). Single values of K Lj were common to all of the relevant data acquired with the same preparation of receptor (i.e. native or alkylated). In the case of QNB, the radioligand and the unlabeled analogue could be assigned single values of K Lj without affecting the sum of squares (i.e., K Pj ϭ K Aj ) (p ϭ 0.37); in the case of NMS, the same constraint compromised the fit (p Ͻ 0.00001). The sum of squares was significantly lower with two classes of sites rather than one (p Ͻ 0.00001), and a single value of F 2 was common to all of the data (F 2 ϭ 0.58 Ϯ 0.06  protein-coupled receptors can interconvert spontaneously between two or more states of different intrinsic activity. Agonists bind with higher affinity to the more active state and shift the equilibrium accordingly, whereas antagonists favor the less active state (e.g. Refs. 15, 16, 36, and 37). The molecular distinction between the two states remains unclear, because the effects of mutation and other perturbations typically are monitored by means of a receptor-elicited response. According to one view, however, the inactive and active states correspond to the free receptor and the receptor-G protein complex, respectively (38). An extension of that view posits that the two states of the receptor differ in their affinity for the ligand on the one hand and the G protein on the other (15,39,40). It also has been suggested that N-ethylmaleimide reacts with muscarinic receptors only in the agonist-specific state, thereby locking the receptor in that conformation and causing an observed increase in the affinity for agonists (10,11).
A model that incorporates the notion of two interconverting states offers a plausible explanation for the effects of N-ethylmaleimide on purified receptors in the present investigation. The lines in Figs. 7 and 9A represent the best fit of Scheme II to the data, taken together, and the parametric values are listed in Tables IV and V. It was assumed that there were two noninterconverting forms of the receptor overall (n ϭ 2). Native receptors were wholly of one form (e.g. Figs. 7, A-C, j ϭ 1, F 2 ϭ 0), and receptors treated with N-ethylmaleimide for 24 h were wholly of the other (e.g. Fig. 7, D-F, j ϭ 2, F 2 ϭ 1). 5 This restriction presupposes that alkylation was complete after 24 h in the absence of N-methylscopolamine, and it follows from the observation that the potency of [ 3 H]quinuclidinyl benzilate was essentially the same after treatment with N-ethylmaleimide for 24 h and 4 days ( Table I). The reaction otherwise was assumed to be incomplete: that is, after treatment for less than 24 h in the absence of N-methylscopolamine or for any time in the presence of N-methylscopolamine (Fig. 9A). Accordingly, those data were analyzed as a mixture comprising both forms of the receptor (i.e. 0 Ͻ F 2 Ͻ 1).
It also was assumed that N-ethylmaleimide was without effect on the affinity of the ligand for either R or R*. All of the data therefore shared single values of K Lj and ␣ Lj for each ligand (i.e. K L1 ϭ K L2 and ␣ L1 ϭ ␣ L2 ). With this constraint and the assumption of homogeneity after alkylation for 24 h, the effects of N-ethylmaleimide are attributed exclusively to a change from K R1 to K R2 ; that is, alkylation is assumed to cause a shift in the distribution of receptors between the two states.
Preliminary analyses with Scheme II indicated that the constraints described above were without effect on the sum of squares, but neither were they sufficient to yield a unique set of parametric values. The ambiguity arises in part from uncertainty over the absolute value of K Rj , particularly after alkylation. The relative value of K Rj is better defined, however, and a lower bound of about 100 was determined by mapping the sum of squares with respect to the ratio of K R1 /K R2 . The value of K R1 is about 0.05-0.1 at all acceptable values of K R1 /K R2 , and the results obtained with K R1 /K R2 fixed at 100 are listed in Tables  IV and V. The good agreement between Scheme II and the data indicates that the model can account for all of the effects illustrated in Figs. 7 and 9A. Based on the fitted value of K R1 , purified M 2 receptors were predominantly but not exclusively in the R* state in the absence of muscarinic ligands (log K R1 ϭ Ϫ1.28, Table V). In the presence of an antagonist, however, the net interconversion associated with comparatively high values of FIG. 9. Effect of N-methylscopolamine and carbachol on the rate of alkylation by N-ethylmaleimide, as monitored by [ 3 H] ␣ L caused the R state to predominate at saturating concentrations of the ligand. Because N-ethylmaleimide decreased K R by 100-fold or more, alkylation caused the R* state to predominate not only in the absence of ligands, precluding even the minor amount of R found with the native preparation, but also at saturating concentrations of antagonist. This difference in the ability of a ligand to promote a redistribution of sites from R* to R accounts for the decrease of at least 10-fold effected by N-ethylmaleimide in the apparent affinity of muscarinic antagonists (Table III).
Because the alkylated receptor was predominantly in the R* state with or without antagonist (Table V), the values of ␣ L K L calculated for antagonists in terms of Scheme II (Table V) agree closely with the corresponding values of EC 50 obtained for alkylated receptors in terms of Scheme I (Table III). In contrast, the values of K L and ␣ L K L from Scheme II (Table V) bracket the corresponding values of EC 50 obtained for the native receptor from Scheme I (Table III) The intermediate values of EC 50 are a consequence of the redistribution of sites from R* to R that occurs upon the binding of the antagonist to the native receptor (Table V).
The lower limit on K R1 /K R2 is related to the marked preference of alkylated receptors for the R* state under all conditions. The extent of the redistribution effected by ligands is governed by ␣ Lj . Antagonists bind more tightly to R (␣ Lj Ͼ 1, Table IV) and therefore promote R over R* (K Rj Ͻ ␣ Lj K Rj , Table V). Because R* predominates with native receptors in the absence of ligand and with alkylated receptors irrespective of ligand (Table V), the value of K R1 must approximate or exceed that of ␣ L2 K R2 . The ratio K R1 /K R2 therefore is limited by ␣ L2 (i.e. ␣ L2 Յ K R1 /K R2 ), or by the single value of ␣ L if N-ethylmaleimide is without effect on the relative affinity of the ligand for R and R* (i.e. ␣ L1 ϭ ␣ L2 ϵ ␣ L ). The fit therefore is compromised at values of K R1 /K R2 that are substantially less than ␣ L ; the value of ␣ L is governed in turn by the difference in EC 50 before and after alkylation.
Because N-ethylmaleimide was without effect on the binding of carbachol and oxotremorine-M, the relative affinity of agonists for R and R* is unclear. This is illustrated in Fig. 11, where the global sum of squares is shown to be independent of ␣ L at lower values and to increase at higher values. The map indicates that the data are consistent with any value of ␣ L smaller than about 10 Ϫ0.2 for carbachol and 10 0.9 for oxotremorine-M. This ambiguity arises from the preponderance of R* under all conditions with respect to the agonist (Table V). Because the system is predominantly in the state potentially of higher affinity for agonists, the distribution of sites between R and R* undergoes little or no change upon addition of the ligand. The state of lower affinity for agonists therefore is unobservable.
Scheme II also can account for the opposing effects of Nmethylscopolamine and N-ethylmaleimide on the affinity of purified receptors for [ 3 H]quinuclidinyl benzilate (Fig. 9A). Native receptor was assumed to be exclusively R 1 or R 1 *, as described above, and the progress of the reaction was modeled as the increase in the fraction of sites identified as R 2 or R 2 * (i.e. Upon treatment with N-ethylmaleimide in the absence of N-methylscopolamine, the fraction F 2 increased from zero initially to 0.72 Ϯ 0.03 after 30 min and to 0.92 Ϯ 0.02 after 4 h; the value after 24 h was set at 1. In the presence of N-[ 3 H]methylscopolamine, the value of F 2 increased to only 0.46 Ϯ 0.03 after 30 min and to 0.70 Ϯ 0.03 after 4 or 24 h. In contrast to N-methylscopolamine, carbachol was without effect on the rate of interconverstion from R 1 to R 2 (Fig.  9B). This difference between N-methylscopolamine and carbachol parallels the different preference of antagonists and agonists for the two states of the receptor. It follows that the antagonist slowed alkylation by favoring a state, in this case the R state, that is comparatively unreactive to N-ethylmaleimide.
Evaluation in Terms of Scheme III-A kinetically determined variant of Scheme II can account for the opposing effects of antagonists and N-ethylmaleimide on the rate of inactivation (Figs. 4 and 5) and, in the case of the alkylated receptor, for the time-dependent, rightward shift in the binding of N-[ 3 H]methylscopolamine (Fig. 3B). Scheme III was formulated with time as an independent variable and incorporates the possibility that the receptor can decay to a state or states that do not bind the radioligand, at least at the concentrations used in the assays. The lines in Figs. 3B and 4 represent the best fit to the pooled data, and the rate constants are listed in the legend to Fig. 4.
Receptors in the state designated as R 1 in Scheme III were stable under the conditions of the assays, with a half-life measured in days either with or without N-[ 3 H]methylscopolamine (i.e. k PR1 and k R1 ). The R 2 state was almost as stable in the presence of antagonist (k PR2 ). In the absence of ligand (k R2 ), however, the decay was 5-fold more rapid with the native receptor and 25-fold more rapid after treatment with N-ethylmaleimide. Alkylation therefore hastened inactivation by further destabilizing the labile R 2 state and by promoting R 2 over the stable R 1 state. The protective effect of N-[ 3 H]methylscopolamine derived in part from its preference for R 1 over R 2 and the attendant redistribution of sites away from the labile state (i.e. ␣ L Ͼ 100, Table IV); also, either state of the receptor was more stable in the presence of a ligand. The latter effect predominated after alkylation, because the value of ␣ L K Rj was such that most of the sites were in the R 2 state even at saturating concentrations of antagonist (Table V). Similarly, occu-pancy per se also accounted for the stabilizing effect of carbachol (Fig. 6).
In the context of Scheme III, the rightward shift shown for N-[ 3 H]methylscopolamine in Fig. 3B derived from the opposing effects of the antagonist and N-ethylmaleimide on the inactivation of alkylated receptors. Because of the decay illustrated in Fig. 4B, longer incubation times increased the concentration of N-[ 3 H]methylscopolamine required to achieve a given level of binding. That led in turn to an increase in the value of EC 50 over time. Scheme III also can account for a decrease in maximal binding, as described below, but the predicted effects differ from those illustrated in Fig. 3B. In terms of the model, the capacity for N-[ 3 H]methylscopolamine was 73-89% of that for [ 3 H]quinuclidinyl benzilate at each time of incubation. The shortfall does not derive from the nonbinding species R 3 and R 4 , because the parameter [R] t comprises all forms of the receptor. Also, the amounts of R 3 and R 4 were assumed to be zero at the outset, and the apparent loss of sites is expected to emerge over time.
The predicted effect of time on the binding of N-[ 3 H]methylscopolamine to the alkylated receptor is illustrated in Fig. 12,

TABLE IV
Effect of N-ethylmaleimide on the affinities of muscarinic ligands for purified M 2 receptor, evaluated in terms of Scheme II The data illustrated in Figs. 7 and 9A were analyzed in concert according to Equations 4 and 6. There were two classes of sites overall (n ϭ 2), representing native receptor (j ϭ 1) and alkylated receptor (j ϭ 2). The reaction with NEM was complete after 24 h, and the data in the lower panels of Fig. 7 therefore were treated as a homogeneous population of sites (j ϭ 2). The data in Fig. 9A were treated as a mixture of sites. Single values of K L , and ␣ L were common to all of the data acquired with both native and alkylated receptor, as described in the text. QNB and NMS were present both as the radioligand (L ϭ P) and as the unlabeled analogue (L ϭ A). The two forms of QNB yielded consistent estimates of K L and ␣ L , which therefore were optimized as a single value (i.e. ␣ P ϭ ␣ A ϵ ␣ L , K P ϭ K A ϵ K L ). The same constraint with NMS led to a significant increase in the sum of squares. Capacities were estimated as the total concentration of sites ([R] t ) and that fraction corresponding to alkylated receptor (i.e. F 2 ϭ ). The number of curves represented in Fig. 7 is shown in parentheses (native receptor, alkylated receptor). Further details are described in the legend to Table V Fig. 7. A separate value was required for the data in Fig. 9A (log K L ϭ Ϫ10.68 Ϯ 0.08), perhaps because a different batch of receptor was used in those experiments.
c The parametric values listed for carbachol and oxotremorine-M correspond in each case to the upper bound on log ␣ L (Fig. 11). Lower values were without effect on the sum of squares or on the values of log K L and log ␣ L obtained for other ligands. The latter were determined with the value of log ␣ L for each agonist fixed at Ϫ2.  Fig. 7 and 9A were analyzed in concert according to Equations 4 and 6. The distribution of sites between R and R* in the absence of ligand was evaluated as the single value of K R2 common to all data acquired with the alkylated receptor and the single value of the ratio K R1 /K R2 common to all data acquired with the native receptor. The value of log (K R1 /K R2 ) was fixed at 2 to obtain the optimized value of log K R2 (Ϫ3.28 Ϯ 0.08) and the corresponding value of log K R1 (Ϫ1.28) listed in the table. The values of log (␣ L K Rj ) were calculated from the value of log K Rj and the appropriate value of log ␣ L from Table IV. The fraction of sites in the R state was calculated from the value of K Rj for the unliganded receptor and from ␣ L K Rj for the receptor at saturating concentrations of the ligand. Further details are described in the legend to Table IV Conformational Status of the Receptor and Activation upon Purification-The present results recall an earlier suggestion that N-ethylmaleimide reacts preferentially with a conformation of the muscarinic receptor that exhibits higher affinity for agonists (10,11). In those studies, however, alkylation was found to increase the affinity of membrane-bound receptors for agonists without affecting that for antagonists; conversely, agonists speeded up the reaction with N-ethylmaleimide, whereas antagonists were without effect. That pattern is opposite to the present observation that alkylation decreased the affinity of purified receptors for antagonists but not agonists while antagonists but not agonists slowed the reaction. In terms of Scheme II, it appears that the membrane-bound receptors were almost exclusively in the R state, and that alkylation led to a decrease in K R . The change caused only a minor redistribution of sites from R to R*, however, and it therefore had no appreciable effect on the binding of antagonists; nonetheless, it was sufficient to permit a major redistribution at saturating concentrations of agonist. The contrasting behavior of membrane-bound and purified receptors suggests that solubilization or purification causes an interconversion from R to R*.
N-Ethylmaleimide generally has been found to have little or no effect on the affinity of antagonists for muscarinic receptors in membranes (e.g. Refs. 8 -10). With agonists, however, alkylation typically leads to an increase in the Hill coefficient and a concomitant decrease in the magnitude of the shift effected by guanyl nucleotides (e.g. Refs. 7-10). The effect on overall affinity has been less consistent, probably because of the complexity of the changes; thus, agonists have been found to bind more tightly in some studies (e.g. Refs. 8,10,13,14) and less so in others (e.g. Refs. 6 and 9). Factors that seem to account for such variability include the temperature of the reaction (41), the concentration of the reagent (9, 10), and the stability of the receptor before and after alkylation. In some cases, the change in the binding of the agonist may arise indirectly from an effect on the G protein (e.g. Ref. 41). N-Ethylmaleimide was without effect on agonists in the present investigation, but the characteristic effects were recovered upon reconstitution of the purified receptor with G o/i . The loss of that sensitivity upon purification therefore was reversible, in accord with the notion that solubilization or purification is accompanied by a decrease in the value of K R . Scheme II or variants thereof can account for the constitutive activity and related effects that have been described for mutants and, in some cases, the native forms of several G proteinlinked receptors (15,16,36,42). The emergence of a common pattern wherein a ligand-regulated equilibrium is affected differently by agonists and antagonists suggests that the states designated here as R and R* represent the inactive and active forms of the M 2 receptor, respectively. That in turn implies that solubilization and purification lead to activation of the receptor. It also suggests that agonists favor R* over R and argues against the alternative possibility that they are indifferent. Wild-type muscarinic receptors in native membranes exhibit some constitutive activity (16,17), but most are presumably in the R state (i.e. K R Ͼ 1). With M 5 receptors transiently expressed in NIH 3T3 cells, for example, only 1.9% of the sites were found to inhabit the R* state in the absence of ligand (16). If antagonists bind preferentially to R (␣ L Ͼ 1), their apparent affinity for membrane-bound receptors, as estimated from Scheme I, is expected to approximate their intrinsic affinity for the R state (EC 50 Ϸ K L ). 6 Purified receptors appear to exist predominantly in the R* state (i.e. K R Ͻ 1); if agonists bind preferentially to R* (␣ L Ͻ 1), their apparent affinity is expected to approximate their intrinsic affinity for the R* state (EC 50 Ϸ ␣ L K L ). It follows that solubilization should be accompanied by a reduction in the apparent affinity of antagonists (K L Ͻ EC 50 Յ ␣ L K L ) and an increase in that of agonists (␣ L K L Ͻ EC 50 Ͻ K L ), assuming that the effects of a decrease in K R are not obscured by concomitant changes in either K L or ␣ L .
Among muscarinic antagonists, solubilization of the receptor generally has been found to reduce affinity or to have little effect (e.g. Refs. [43][44][45]. Antagonists therefore behave as expected if detergents were to decrease K R by varying degrees, depending upon the conditions. Detergents and N-ethylmaleimide also may have a cumulative effect, as suggested by the early observation that N-ethylmaleimide reduced the binding of [ 3 H]quinuclidinyl benzilate only when pretreated membranes from porcine brain were solubilized in L-␣-lysophosphatidylcholine or when alkylation was carried out on the solubilized receptor (46). Solubilization has been reported to increase the affinity of the M 1 -selective antagonist pirenzepine for M 2 receptors (45,47), but such examples are rare and perhaps highly specific.
Muscarinic agonists generally bind with lower affinity after solubilization (48), in contrast to predictions based narrowly on Scheme II, but the decrease may be dominated by changes in the interaction between the receptor and the G protein. G proteins do not appear explicitly in Scheme II, and the nature of their involvement is unclear. Also, detergents have been shown to reduce the potency of both agonists and antagonists when purified M 2 receptors were compared in solution and after reconstitution in phospholipid vesicles (49). The solubilization of porcine brain in cholate was found to reduce the potency of eight muscarinic antagonists with little or no change in that of five agonists (44). Such observations suggest that solubilization may effect more than a decrease in K R : for example, a concomitant increase in K L would potentiate the increase in EC 50 expected with antagonists but tend to offset the decrease expected with agonists.
In accord with the predictions of Scheme II, the affinity of isoproterenol for the ␤ 2 adrenergic receptor and a constitutively active mutant was increased as expected if solubilization and purification were to cause a shift from the inactive to the active state (15). Also, the suggestion that N-ethylmaleimide destabilizes the M 2 receptor by favoring the active state is consistent with other evidence that activated G protein-coupled receptors are comparatively unstable. Constitutively active mutants of both the ␤ 2 adrenergic receptor (15,19) and the ␣ 2A adrenergic receptor (20,50) have been shown to undergo spontaneous inactivation more rapidly than the corresponding wildtype receptors.
Effects Inconsistent with Schemes I-III-A final comment is in order concerning two anomalies that cannot be accommodated by either Scheme I or Schemes II and III. With native and alkylated receptors, the affinity of N-[ 3 H]methylscopolamine in terms of Scheme I exceeded that of the unlabeled analogue. The difference was only about 1.6-fold with native receptors but increased to 3-7-fold after treatment with Nethylmaleimide (Table III). Also, the relative capacity for N-[ 3 H]methylscopolamine and [ 3 H]quinuclidinyl benzilate was greater than 0.9 with native receptors but only about 0.77 after alkylation (Fig. 3B, Table II). Similar discrepancies emerge in terms of Schemes II and III. Labeled and unlabeled N-methylscopolamine revealed a difference of 1.5-and 7.4fold in the value of K L and ␣ L K L , respectively (Table V), and the relative value of [R] t inferred for N-[ 3 H]methylscopolamine and [ 3 H]quinuclidinyl benzilate after alkylation was about 0.8 (Fig. 3B).
Because there was no discrepancy in the affinity of quinuclidinyl benzilate, the effect with N-methylscopolamine seems to be ligand-specific. It has been noted previously that N-[ 3 H]methylscopolamine can be contaminated with scopolamine (21), which would yield artifactually low estimates of both the capacity and the dissociation constant, but scopolamine was absent from the product used in the present investigation. Triethylamine was identified in the mass spectrum, but its equilibrium dissociation constant at M 2 receptors exceeds 0.1 mM (51). Because the molar ratio of the impurity to N-[ 3 H] methylscopolamine never exceeded 0.7, any effect on binding was negligible (21).
The appearance of both anomalies after treatment with N-ethylmaleimide raises the possibility that they derive from the reduced stability of alkylated receptors. That seems unlikely, however, because the shortfall in capacity for N-[ 3 H]methylscopolamine cannot be accounted for in terms of Scheme III. Also, the apparent affinity of N-[ 3 H]methylscopolamine exceeded that of the unlabeled analogue (Table III). In the event of instability, the affinity of unlabeled N-methylscopolamine inferred from its inhibitory effect at near saturating concentrations of [ 3 H]quinuclidinyl benzilate should exceed the apparent but artifactually low affinity of the radioligand. The expected effect of instability on the binding of N-[ 3 H]methylscopolamine can be seen in Fig. 3B, where the profile shifts rightward over time. Even after 3 h, however, the radioligand remained at least as potent as the unlabeled ligand (cf . Tables II and III).
The difference in the capacity of alkylated receptors for [ 3 H]quinuclidinyl benzilate and N-[ 3 H]methylscopolamine implies that about 20% of the sites are of anomalously low affinity for the latter. No corresponding shoulder is evident in the inhibitory profile of unlabeled N-methylscopolamine, which therefore appears to inhibit at sites to which N-[ 3 H]methylscopolamine does not bind. Essentially the same pattern has been described previously for M 2 muscarinic and D 2 dopaminergic receptors, and the apparent paradox could be rationalized in terms of cooperative interactions between the radiolabeled and unlabeled ligands (22,51,52). Similar effects may account for the difference in the capacity of alkylated receptors for N-[ 3 H]methylscopolamine and [ 3 H]quinuclidinyl benzilate in the present investigation.