Structural instability of a constitutively active G protein-coupled receptor. Agonist-independent activation due to conformational flexibility.

Mutations in several domains can lead to agonist-independent, constitutive activation of G protein-coupled receptors. However, the nature of the structural and molecular changes that constitutively turn on a G protein-coupled receptor remains unknown. Here we show evidence that a constitutively activated mutant of the β2 adrenergic receptor (CAM) is characterized by structural instability and an exaggerated conformational response to ligand binding. The structural instability of CAM could be demonstrated by a 4-fold increase in the rate of denaturation of purified receptor at 37°C as compared with the wild type receptor. Spectroscopic analysis of purified CAM labeled with the conformationally sensitive and cysteine-reactive fluorophore, N,N′dimethyl-N-(iodoacetyl)-N′-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)ethylenediamine, further indicated that both agonist and antagonist elicit more profound structural changes in CAM than in the wild type protein. We propose that the mutation that confers constitutive activity to the β2 adrenergic receptor removes some stabilizing conformational constraints, allowing CAM to more readily undergo transitions between the inactive and the active states and making the receptor more susceptible to denaturation.

In classical receptor theory, binding of agonist has been considered essential for receptor activation and transmission of the biological signal across the plasma membrane. However, expression of many members of the seven-transmembrane, G protein-coupled receptor family in transfected cell lines has revealed that activation of intracellular messenger systems can occur in the absence of agonists (1)(2)(3)(4)(5)(6). Moreover, it has been found that mutations in several different regions of G proteincoupled receptors can dramatically enhance agonist-independent receptor activity and in some cases confer oncogenic properties to the receptor (1)(2)(3)(7)(8)(9)(10)(11)(12)(13)(14). Nevertheless, the molecular and structural changes in the receptor that are responsible for constitutive, agonist-independent activation are poorly understood. According to the prevailing two-state model for activation of G protein-coupled receptors, constitutive activation has been explained as a disturbance of the normal equilibrium between the inactive (R) 1 state and the active (R*) state leading to a higher proportion of receptor molecules in the active R* state (3,15). However, the conformational state of a constitutively active receptor protein has so far only been deduced from its effects on intracellular second messenger systems; therefore, this hypothesis has not been substantiated by any direct structural analysis.
Recently, we have described the use of fluorescence spectroscopy to directly analyze ligand-induced conformational changes in the purified wild type ␤ 2 adrenergic receptor (16). The approach is based on the sensitivity of many fluorescent molecules to the polarity of their molecular environment (17). In this study, we use the same techniques to directly study conformational changes associated with constitutive activation. Our results reveal novel characteristics of a constitutively active receptor that provide insight into the mechanism of altered signaling behavior of this mutant ␤ 2 receptor.

EXPERIMENTAL PROCEDURES
Expression of the Constitutively Activated ␤ 2 Receptor in Insect Cells-The cDNA encoding CAM was generously provided by Dr. R. J. Lefkowitz (Duke University, Durham, NC) and cloned into the baculovirus expression vector pVL1392 (Invitrogen) as described for the wild type receptor (18). The resulting construct had the cleavable influenzahemagglutinin signal-sequence followed by the FLAG epitope (IBI) at the amino terminus and a tail of six histidines at the carboxyl terminus. Baculovirus containing the tagged CAM sequence was generated using the BaculoGold kit (Pharmingen) and plaque-purified. Sf-9 cells were grown as described (16).
Assessment of Receptor Expression in the Presence and the Absence of Ligand-Suspension cultures of Sf-9 cells at a density of 3 ϫ 10 6 cells/ml were infected in the presence or the absence of ligands plus 2 mM ascorbate. After 48 h 1.0-ml samples of the cultures were collected, and ligands were added to controls before centrifugation (10 min at 10,000 ϫ g). The pellets were frozen at Ϫ70°C, thawed, and resuspended in 1.0 ml of lysis buffer (10 mM Tris-HCl, pH 7.4, with 1 mM EDTA, 10 g/ml leupeptin, 10 mg/ml benzamidine, and 0.2 mM phenylmethylsulfonyl fluoride). The lysed cells were centrifuged (10 min at 10,000 ϫ g), and the membrane pellet was washed three times in 1.0 ml of binding buffer (75 mM Tris-HCl, pH 7.4, with 12.5 mM MgCl 2 , and 1 mM EDTA) plus 10 g/ml leupeptin and 10 g/ml benzamidine. For binding assay 50 l of membrane suspension was incubated in binding buffer in a total volume of 0.5 ml with 10 nM [ 3 H]dihydroalprenolol (Amersham Corp.) for 2 h as described (19).
Purification Procedure-For purification, Sf-9 cells were grown in 600-ml cultures, infected with a 1:100 dilution of a high titer virus stock at a density of 3 ϫ 10 6 cells/ml and harvested after 48 h. To avoid a complete loss of the unstable CAM protein it was necessary to exclude the M1 anti-FLAG antibody column from our previously described purification procedure allowing purification in 1 day instead of over 2 days (16,20). Briefly, lysed cell membranes were solubilized in 1.0% * This work was supported in part by National Institute of Health Grants RO 1 NS28471, DA-09083, and DA-00060 and by the Harold G. and Leila Y. Mathers Charitable Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
(w/v) n-dodecyl-␤-D-maltoside (D␤M) (CalBiochem) followed by nickel column chromatography using Chelating Sepharose (Pharmacia Biotech Inc.) and alprenolol affinity chromatography as described (16,20). 1-2 nmol of pure CAM protein generally could be obtained from three 600-ml cultures. The specific activity of purified CAM was ϳ3 nmol/mg protein and ϳ8 nmol/mg protein for the WT purified by the same procedure. Purified protein was analyzed by 10% SDS-polyacrylamide gel electrophoresis. The protein was either visualized by Coomassie staining or by Western blotting using the M1 anti-FLAG antibody (IBI) and the ECL system (Amersham Corp.).
Assessment of Purified Receptor Stability-Nickel-purified receptor (2 pmol) was incubated at 37°C for up to 3 h in 100 l of 100 mM Tris-HCl, pH 7.5, containing 0.08% (w/v) D␤M, 200 mM NaCl plus 4 mg/ml bovine serum albumin. The amount of functional receptor was determined in binding assays using a saturating concentration of [ 3 H]dihydroalprenolol (10 nM) according to described methods (16). To investigate the ability of ligands to protect the receptor from degradation, 2 pmol of CAM was incubated for 15 min and 2 pmol of WT was incubated for 60 min at 37°C in the above mentioned buffer plus 2 mM ascorbate with our without ligands. Ligands were removed by binding receptors to 100-l nickel columns (chelating Sepharose) that were subsequently washed 8 -10 times with 500 l of buffer. Receptors were eluted in 500 l of buffer with 200 mM imidazol before binding analysis.
Fluorescence Labeling and Spectroscopy-Purified receptor was bound to a 150-l nickel column (chelating Sepharose). Labeling with the cysteine-reactive fluorophore IANBD (Molecular Probes, Eugene, OR) was achieved by recycling 1.0 ml of 0.5 mM IANBD in buffer (Tris-HCl, pH 7.5, containing 500 mM NaCl, and 0.08% D␤M) with 10 Ϫ6 M alprenolol over the nickel column for 20 min. Excess dye was removed by extensive washing of the column with approximately 50 column volumes of buffer. Labeled receptor was eluted with 200 mM imidazol in buffer. The labeling procedure resulted in incorporation of 1.0 -1.5 mol of IANBD/mol of WT or CAM receptor, as determined by measuring absorption at 481 nm and using an extinction coefficient of 21,000 M Ϫ1 cm Ϫ1 for IANBD and a molecular weight of 50,000 for the receptor. Protein concentration was determined using the Bio-Rad DC protein assay kit. The results with IANBD-labeled wild type receptor purified and labeled by the procedures described here were indistinguishable from results obtained using our previous procedures (16). Fluorescence spectroscopy was performed at room temperature on a SPEX Fluoromax spectrofluorometer with photon counting as described previously (16).

RESULTS AND DISCUSSION
Lefkowitz and co-workers previously showed that a discrete change in the carboxyl-terminal part of the third intracellular loop leads to constitutive activation of the ␤ 2 adrenoreceptor (CAM) (3). We expressed an epitope-tagged version of this mutant in Sf-9 insect cells to obtain large quantities of receptor needed for purification and subsequent structural characterization. The functional properties of CAM in insect cell membranes were similar to those observed for CAM in membranes from transfected mammalian cells (3). As expected we observed increased agonist affinity, an elevated basal level of adenylyl cyclase and GTPase activity, and a higher maximal agoniststimulated adenylyl cyclase and GTPase activity for CAM than for the ␤ 2 WT receptor when expressed at a similar level (data not shown). Together this confirms that CAM not only possesses constitutive activity but also is "superactive" compared with the wild type receptor.
The expression of functional CAM in Sf-9 insect cells was considerably lower than for the wild type (3.4 Ϯ 1.3 pmol/mg protein versus 8.7 Ϯ 1.8 pmol/mg protein, mean Ϯ S.E., n ϭ 4) in agreement with earlier studies in mammalian cells (3). However, incubation of the cells with either an agonist or an inverse agonist (also referred to as negative antagonist) during the 48-h infection markedly increased the expression of CAM (Fig.  1). A similar increase was also observed for neutral antagonists (data not shown). This surprising lack of correlation between the increase in expression and the pharmacological properties of the added ligands strongly argues that the lowered expression of CAM in the insect cells cannot be explained by receptor down-regulation as a consequence of constitutive activation.
Rather, the increased expression of CAM in the presence of ligand may be due to biochemical stabilization of an inherently unstable protein. Of interest, an even more dramatic up-regulation is observed in transgenic mice expressing CAM in response to antagonist treatment (21). Most likely, these data also reflect ligand stabilization of an inherently labile protein.
The instability of CAM can be demonstrated by observing the rate of denaturation at 37°C as shown in Fig. 2A. Assuming an exponential decay, t1 ⁄2 for degradation of CAM was 12.3 min versus 49.9 min for the wild type receptor ( Fig. 2A). The decrease in binding activity for both wild type and CAM could be partially prevented by both the agonist, isoproterenol, and the inverse agonist, ICI 118,551 (22,23) with isoproterenol being slightly less effective than ICI 118,551 (Fig. 2B). Western blotting of partially purified receptor before and after exposure to 37°C for 3 h revealed essentially no changes in the intensity of the receptor band and no evidence of proteolysis ( Fig. 2A, inset).
The cysteine-reactive fluorescent probe IANBD can be used as a sensitive molecular reporter of ligand-induced conformational changes in the ␤ 2 adrenoreceptor (16). Agonist stimulation of purified IANBD-labeled ␤ 2 receptor leads to a dose-dependent, reversible decrease in fluorescence, indicating that one or more cysteines labeled with IANBD are exposed to a slightly more polar environment upon agonist binding. By systematic mutation of cysteines we have localized the responsible residues to the third and sixth transmembrane domain. 2 There was a linear correlation between the magnitude of the fluorescence change in response to a series of adrenergic agonists and the intrinsic efficacy of the compounds in adenylyl cyclase assays implying that the agonist-mediated decrease in fluorescence ensues from a conformational change involved in receptor activation and G protein coupling (16). In contrast to agonists, we found that inverse agonists caused a relative increase in fluorescence emission (16). Emission scans of IANBD-labeled CAM and wild type receptor both showed maximal fluorescence at an emission wavelength of 523-526 nm (data not shown). Time course analysis revealed that stimulation of IANBD-labeled CAM with the full agonist, isoproterenol, and the partial agonist, salbutamol, elicited substantially greater decreases in fluorescence emission than in the IANBD-labeled WT receptor (Fig. 3, A-E). In addition, the ratio of the salbutamol response relative to the isoproterenol response increased from 0.40 in the WT to 0.76 in CAM (Fig. 3E). This agrees with 2 U. Gether, S. Lin, and B.K. Kobilka, unpublished observation. the well described increase in the efficacy of partial agonists at constitutively activated receptor mutants in biological assays (Refs. 3 and 4 and data not shown). The changes were fully reversible by antagonist for both receptors (Fig. 3, A-D). As with the WT receptor, the inverse agonist ICI 118,551 induced an increase in fluorescence of the IANBD-labeled CAM receptor; however, the magnitude of the response was greater for the CAM receptor (Fig. 3E). The larger changes in fluorescence induced by both agonists and inverse agonists in the IANBDlabeled CAM receptor are even more impressive considering that the specific activity of CAM was lower than that of the WT (CAM, 3 nmol/mg protein; WT, 8 nmol/mg protein).
Taken together, our findings delineate two novel properties of a constitutively activated receptor: structural instability and an exaggerated conformational response to drug binding. Our findings, in particular the larger changes in fluorescence observed with CAM in response to agonists, are unexpected. If we assume that the CAM mutation simply causes the unliganded receptor to adopt a conformation mimicking the agonist bound form of the WT receptor, then the effect of agonist binding might have been expected to produce little additional change in fluorescence. One likely explanation for this apparent discrepancy could be that constitutive activation of the ␤ 2 receptor confers a higher degree of conformational flexibility to the receptor protein due to the disruption of stabilizing conformational constraints. This higher degree of conformational flexibility may allow CAM to more readily undergo transitions between the R state and the R state* in response to ligand binding, thus leading to larger fluorescence changes and structural instability (as outlined below and in the model in Fig. 4). In this thermodynamic model, we propose that the R* state in the prevailing two-state model for activation of G protein-coupled receptors (3,(23)(24)(25)(26)(27) can be considered a high energy, intermediate state that can be stabilized by the G protein and/or the agonist (Fig. 4). However, in the absence of agonist or G protein, the expected lifetime of the excited R* state for both CAM and the WT receptor would be short due to its high energy. Receptor in R* will either rapidly return to the ground state or denature. Therefore, at any given time the fraction of both CAM and WT receptor in R* should be rather low in absence of agonist and G protein ([R* WT ] ϽϽϽ [R WT ] and [R* CAM ] Ͻ Ͻ [R CAM ]) (Fig. 4). Our model proposes that agonists stabilize both R* WT and R* CAM ; however, because of the smaller energy difference between the R and R* state of CAM, agonists would cause a higher proportion of CAM molecules than WT molecules to undergo the transition from R to R*. For example, agonist occupancy of the wild type receptor in the absence of G protein might cause a change in the R* WT fraction from 5 to 20%, whereas the change could be from 10 to 35% for CAM. This greater change in the fraction of R* for CAM than for the wild type receptor in response to agonists is consistent with the larger changes in fluorescence, because the change in fluorescence most likely reflects receptor molecules undergoing transition from R to R* (16). Moreover, it explains the superactivity of agonist-activated CAM as compared with agonistactivated WT measured in GTPase and adenylyl cyclase assays (3). Finally, the model can explain the larger increase in fluorescence over time in response to the inverse agonist ICI  (C and D). The responses were reversed by the neutral antagonist alprenolol (10 Ϫ3 M), which has a more than 100-fold higher affinity for the receptors than isoproterenol and salbutamol (data not shown). By itself, alprenolol did cause a relative increase in baseline fluorescence over time but with a slower time course than the reversal of the agonist response (data not shown). The ligand concentrations used were chosen to ensure saturation of the receptors eliminating any influence from different agonist affinities. Excitation was at 481 nm, and emission was measured at 525 nm. 118,551. ICI 118,551 stabilizes the receptor in the ground state R and, as discussed above, the fraction of CAM molecules in R* would always be higher than for the WT receptor in the absence of ligands; therefore, the population of receptor going from R* to R upon the addition of the inverse agonist would be greater, and a larger increase in fluorescence would be expected.
An alternative explanation for our observations would be that CAM has an altered structure with an altered environment around the fluorophore, and the active state (R* CAM ) is structurally distinct from the active state of the WT (R* WT ). Thus, CAM may accommodate a superactivated state that interacts more efficiently with the G protein and causes the more dramatic changes in fluorescence. Our data cannot distinguish between these two explanations. However, the model we have outlined in Fig. 4 represents the simplest hypothesis, because it does not propose any changes in the tertiary structure of R* for CAM as compared with R* for the WT receptor. Thus it is compatible with the prevailing two-state model for activation of G protein-coupled receptors (3,(23)(24)(25)(26)(27). It is also highly unlikely that the environment around the fluorophore is markedly changed in CAM, because the mutant has the same stoichiometry of IANBD labeling (see "Experimental Procedures") and the emission maxima from IANBD-labeled CAM and WT are indistinguishable (data not shown).
The structural instability should also be expected according to the model outlined in Fig. 4. Due to the higher energy, the excited R* state is predicted to be structurally more unstable; therefore it follows that CAM is more susceptible to denatur-ation. A higher structural instability of the R* state as compared with R is supported by molecular dynamics simulations of ligand-receptor complexes (28 -30). CAM may also be more susceptible to denaturation from the ground state due to the proposed higher energy level of this state (R CAM ). Of interest, the association of structural instability with constitutive activation is not restricted to G protein-coupled receptors. A constitutively active mutant of the Gs␣ protein has been shown to be thermolabile and to degrade rapidly at 37°C (31). Analogous to the observed ligand stabilization of CAM, the degradation of the constitutively active Gs 224 protein could be prevented by the addition of nucleotides (31).
In conclusion, this study represents the first direct structural analysis of a constitutively active G protein-coupled receptor. Interestingly, the data can be predicted from an activation model of G protein-coupled receptors that involves formation of an unstable high energy intermediate, which can be considered analogous to the active R* state in the prevailing two-state model for activation of G protein-coupled receptors (3,(23)(24)(25)(26)(27). Hence, our results underscore the importance of constitutively active mutants as critical tools for understanding the molecular processes involved in activation of G protein-coupled receptors.
FIG . 4. A simplified, thermodynamic energy state diagram for activation of G protein-coupled receptors. The diagram represents a system at equilibrium. The abundance of any single state at equilibrium is determined by its relative energy. Receptor activity for both WT and the CAM receptor is determined by the ability of the receptor to overcome the energy difference between the R and R* states. The relative energies among the receptor states in this qualitative energy diagram can be inferred from known properties of the receptor. First, the energy difference between R and R* is smaller for CAM than for WT. This is indicated by the dramatic increase in basal and agonist induced activity of CAM compared with WT, which implies that it is energetically easier to reach R* CAM from R CAM than it is to reach R* WT from R WT. Second, the energy of an active R* state (in the absence of agonist or G protein) is higher than that of the R state. This is indicated by the low basal activity levels of WT and the ability of agonists (A) to increase the population of activated receptors for both WT and CAM. Third, for the agonist-bound receptor, the energy of an active receptor bound to agonist (AR*) (in the absence of G protein) is higher than for the inactive receptor bound to agonist (AR). This inference comes from the known effect of GTP on agonist affinity in the presence of G protein. GTP abolishes (or significantly decreases) high affinity binding by reducing G protein coupling; thus the high affinity binding component is taken to represent AR*. Our studies are done in the absence of G protein, and only low affinity agonist binding is observed suggesting that the AR* state is sparsely populated relative to AR and thus must be of higher energy. As discussed in the text, the smaller energy difference between AR and AR* for CAM as compared with the WT receptor predicts a larger change in the fraction of CAM in R* following exposure to agonist and thus a larger change in fluorescence.