The effect of ligand efficacy on the formation and stability of a GPCR-G protein complex

Abstract

G protein-coupled receptors (GPCRs) mediate the majority of physiologic responses to hormones and neurotransmitters. However, many GPCRs exhibit varying degrees of agonist-independent G protein activation. This phenomenon is referred to as basal or constitutive activity. For many of these GPCRs, drugs classified as inverse agonists can suppress basal activity. There is a growing body of evidence that basal activity is physiologically relevant, and the ability of a drug to inhibit basal activity may influence its therapeutic properties. However, the molecular mechanism for basal activation and inhibition of basal activity by inverse agonists is poorly understood and difficult to study, because the basally active state is short-lived and represents a minor fraction of receptor conformations. Here, we investigate basal activation of the G protein Gs by the β₂ adrenergic receptor (β₂AR) by using purified receptor reconstituted into recombinant HDL particles with a stoichiometric excess of Gs. The β₂AR is site-specifically labeled with a small, environmentally sensitive fluorophore enabling direct monitoring of agonist- and Gs-induced conformational changes. In the absence of an agonist, the β₂AR and Gs can be trapped in a complex by enzymatic depletion of guanine nucleotides. Formation of the complex is enhanced by the agonist isoproterenol, and it rapidly dissociates on exposure to concentrations of GTP and GDP found in the cytoplasm. The inverse agonist ICI prevents formation of the β₂AR-Gs complex, but has little effect on preformed complexes. These results provide insights into G protein-induced conformational changes in the β₂AR and the structural basis for ligand efficacy.

G protein-independent signaling pathways have been identified for a number of G protein-coupled receptors (GPCRs) (1–3); however, GPCR–G protein interactions represent the fundamental signaling interface, underlying the physiologic response to the majority of hormones and many neurotransmitters. Crystal structures of G proteins have been obtained in both active and inactive states (4–6), and structures of bovine rhodopsin (7–12), squid rhodopsin (13, 14), bovine opsin (15, 16), the human β₂AR (17–19), the turkey β₁AR (20), and the adenosine A2a receptor (21) have been reported. However, relatively little is known about the active-state GPCR-G protein complex.

The active-state GPCR-G protein complex will be defined as the physical association between a GPCR and a G protein that promotes GTP binding to the Gα subunit. This complex can be identified by the specific allosteric effects that each protein imposes on the other. Agonist-bound GPCRs promote exchange of GDP for GTP on the Gα subunit. G protein effects on receptor structure are more difficult to detect, but agonist binding affinity for many GPCRs is enhanced when complexed with a G protein. This effect was originally demonstrated for the β₂AR (22) and led to the ternary complex model of receptor activation (23). The enhanced agonist affinity observed for the active-state complex is abolished by both GTP and GDP suggesting that the bound G protein is nucleotide-free (24).

The active-state complex should be distinguished from other types of physical association or co-localization of GPCRs and G proteins that may be observed in cells and may be important for signal transduction. Recently it has been possible to study GPCR–G protein interactions in cells by using FRET between components tagged with fluorescent and/or luminescent proteins (25–27). These studies provide evidence that a fraction of receptors and G proteins may exist in a precoupled or preassociated state. These associations may occur through direct receptor–G protein interactions, or through interactions with common scaffolding proteins. However, this precoupled or preassociated state probably differs from the active-state or ternary complex as defined above, because G protein-dependent high-affinity agonist binding is not observed at GTP and GDP concentrations present in intact cells (28, 29). Thus, in the presence of cytosolic guanine nucleotides, the active-state complex would be short-lived.

For many GPCRs, the active-state complex can form in the absence of agonists leading to a certain level of basal, agonist-independent activity also called constitutive activity. There is a growing appreciation that drugs that inhibit basal activity, called inverse agonists, may be more effective therapeutics for some indications than neutral antagonists—drugs that competitively inhibit agonist binding but have little or no effect on basal activity (30).

Little is known about the mechanism for basal activity and inverse agonism. Here, we use site-specific labeling of β₂AR with a conformationally sensitive fluorescent probe together with recombinant HDL particles (31), a highly efficient reconstitution system, to investigate the active-state complex. This experimental system allows us to directly monitor ligand- and G protein-induced conformational changes in the β₂AR. We observe that monomeric β₂AR forms a stable complex with the Gs heterotrimer in the absence of guanine nucleotides. This complex can be detected by changes in the fluorescence of labeled β₂AR. Conformational changes induced in the β₂AR by Gs alone or by agonist alone result in similar changes in the fluorescence of labeled β₂AR. The β₂AR-Gs complex rapidly dissociates in the presence of both GDP and GTP. A saturating concentration of the inverse agonist ICI-118,551 does not disrupt the preformed complex, but prevents complex formation. In contrast, the neutral antagonist alprenolol has little effect on the stability of the complex nor does it prevent complex formation. These results provide insights into G protein-induced changes in β₂AR structure and the effect of ligand efficacy on the formation and stability of the β₂AR-Gs complex.

Results and Discussion

Site-Specific Labeling of β₂AR with Monobromobimane, an Environmentally Sensitive Fluorophore.

To detect agonist and G protein-induced conformational changes in the β₂AR, we introduced a conformationally sensitive fluorophore adjacent to the G protein-coupling region of transmembrane segment (TM) 6. We previously showed that fluorophores covalently bound to C265 at the cytoplasmic end of TM6 (Fig. 1) are capable of detecting agonist-induced conformational changes (32–34). C265 is well positioned to detect conformational changes associated with G protein activation. The recent crystal structure of the human β₂AR (17) predicts that monobromobimane bound to C265 lies in a relatively nonpolar pocket formed at the interface between TMs 3, 5 and 6 (Fig. 1C). The recent structure of opsin bound to the carboxyl terminus of the Gα-subunit of transducin (16) shows that the G protein interacts with residues in the inner side of the cytoplasmic TM5 and TM6. A similar pattern of interaction between the β₂AR and Gs would involve residues V222^5.61, A226^5.65, and Q229^5.68 in TM5, and K263^6.25, F264^6.26, K267^6.29, and A271^6.33 in TM6, in the vicinity of C265^6.27 (see Fig. 1D). Conformational changes associated with activation of the β₂AR would be expected to displace bimane bound to C265 to a more polar environment.

Fig. 1.

Site-specific labeling of purified β₂AR with monobromobimane. (A) Sequence and secondary structure of the human β₂AR showing sites where reactive cysteines were mutated to the indicated amino acid (black circles with white letters). C265 is shown in red. The remaining cysteines (gray circles) are not reactive toward monobromobimane. (B) The structure of bimane covalently bound to C265 after reaction with monobromobimane. (C) In the inactive structure of the β₂AR, bimane bound to C265 is predicted to occupy a cavity between TM3, TM5, and TM6. When TM6 adopts an active conformation (because of agonist binding or due to constitutive activity), bimane is displaced (black arrow) out of this cavity to a more polar environment, which is detected as a change in fluorescence intensity and λ_MAX. (D) A model of the active state of the β₂AR in complex with the carboxyl terminal peptide of Gα_S, based on the crystal structure of opsin in complex with a transducin peptide. The amino acids at the positions marked by solid spheres are predicted to form interactions with Gα_S. The residues of the β₂AR are numbered according to their position in the sequence followed by the Ballesteros general number (49) in superscript. In this numbering scheme, the most conserved residue within each helix is designated x.50, where x is the number of the transmembrane helix. All other residues on that helix are numbered relative to this conserved position. The model of the active conformation of β₂AR shown in Fig. 1D was built by homology modeling using the β₂AR (2RH1) and opsin (3CAP) structures as templates. The Gα_s fragment bound to the active-like β₂AR was modeled by threading the 11 C-terminal residues of Gα_s on the structure of the synthetic peptide derived from the carboxy terminus of Gα_t bound to opsin. The surface of the cytoplasmic side of the receptor shown in Fig. 1C was calculated with CastP (50). Homology models and figures were prepared with PyMOL (DeLano WL (2002) The PyMOL Molecular Graphics System. Available at http://www.pymol.org).

C265 is highly reactive to labeling with polar, cysteine-reactive fluorophores (19, 32, 35). The remaining reactive cysteines can be removed by mutagenesis (C77V, C327S, C378A, and C406A) without altering receptor function (35) (Fig. 1A). We previously found that the receptor palmitoylation site (C341) is not reactive (32). Stable palmitoylation of C341 was confirmed in the crystal structure of the β₂AR where protein was purified and alkylated with 4 mM iodoacetamide (100-fold molar excess) under the same conditions. Both the palmitate bound to C341 and iodoacetamide bound to C265 were clearly ordered in the crystal structure.

The modified receptor was expressed in Sf9 insect cells by using recombinant baculovirus technology and purified as previously described (35). Purified β₂AR was labeled with an equivalent amount of monobromobimane. This modified β₂AR labeled at Cys-265 with monobromobimane will be referred to as mB-β₂AR. Monobromobimane is an ideal fluorophore for these experiments because its small size (approximately equivalent to tryptophan, Fig. 1B) and short linker, together with its high sensitivity to the polarity of its molecular environment. These properties make it highly sensitive for detecting conformational changes without interfering with receptor function. Any ligand or G protein-induced movement at the cytoplasmic end of TMs 3, 5, or 6 would be expected to change the molecular environment, and therefore the fluorescence properties, of bimane bound to C265.

Agonist- and Gs-Induced Changes in mB-β₂AR Reconstituted into Recombinant HDL Particles.

β₂AR requires a lipid bilayer to efficiently couple to Gs. Whorton et al. recently showed that purified β₂AR can be reconstituted into recombinant HDL particles (rHDL) as monomers, and that monomeric β₂AR couples efficiently to Gs (31). Thus, mB-β₂AR was reconstituted into rHDL, and the response to the agonist isoproterenol (ISO) was determined. Fig. 2A shows the emission spectra of mB-β₂AR in the absence of G protein but in the presence of increasing concentrations of the full agonist ISO. We observe a dose-dependent decrease in fluorescence intensity and an increase in the maximal emission wavelength (λ_MAX) of 15 nm (Fig. 2A). The concentration-dependent effects of ISO on intensity and λ_MAX are shown in supporting information (SI) Fig. S1 A and B. It should be noted that the EC₅₀ for the fluorescence dose–response curve (Fig. S1A, blue curve) is approximately 3-fold higher than the IC₅₀ for ISO in a conventional competition binding experiment (Fig. 2D, black curve). This difference may in part be due to the fact that conformational assays were performed at much higher receptor concentrations (50–100 nM) than were used for conventional ligand binding experiments. Consequently, responses to lower concentrations of ligand may be influenced by ligand depletion. It is also possible that an agonist binding event is not always associated with a conformational change. Agonists such as ISO have relatively low affinity and very rapid on and off rates. ISO can occupy the binding pocket for long enough time to compete with a radiolabeled antagonist, but this binding event may not always be associated with an activating conformational change.

Fig. 2.

Reconstitution of purified bimane-labeled β₂AR (mB-β₂AR) and Gs into rHDL particles. (A) Bimane emission spectra of reconstituted mB-β₂AR in the absence (black spectrum) and presence of increasing concentrations of the agonist ISO (red spectra). Agonist-induced conformational changes lead to a decrease in fluorescence intensity of bimane and a shift in the λ_MAX. (B and C) Bimane emission spectra of mB-β₂AR reconstituted with Gs in the absence of agonist or nucleotides (black spectrum) and in the presence of increasing concentrations of GTPγS (B) and GDP (C) (red spectra). All spectra were normalized to the unliganded, uncoupled state of mB-β₂AR. For experiments in A, this state was the value for the sample without ligand. For experiments in B and C, this state was the value obtained following the addition of 10 μM GTPγS. Each spectrum in A–C represents the average of 3 independent experiments. (D) ISO inhibition of [³H]DHAP binding to mB-β₂AR in rHDL in the presence (red) and absence (black) of Gs. Inset shows the effect of 10 μM GTPγS on ISO bind affinity to mB-β₂AR reconstituted with Gs. These data were fit by using Prism 5.0 (Graphpad). (E) Time-scan of fluorescence monitored at 450 nm. The effect of 200 nM GTP, GTPγS, and GDP on mB-β₂AR-Gs are compared. GTP induces dissociation of mB-β₂AR and Gs followed by reformation of the complex after GTP hydrolysis.

Purified Gs heterotrimer (Gα_Sβ₁γ₂) was added to rHDL containing mB-β₂AR at a ratio of 10 Gs per β₂AR monomer to ensure that mB-β₂AR would have access to at least 1 Gs trimer [during the reconstitution, some Gs is lost because of aggregation due to removal of detergent (31)]. Gs induces a high affinity state for the agonist ISO (K_i 1.6 nM in the presence of Gs compared with 840 nM in the presence of Gs and GTPγS, or mB-β₂AR alone, Fig. 2D) comparable to what is observed in unlabeled β₂AR (31). Thus, modification of C265 with bimane does not interfere with G protein coupling. Using these reconstitution conditions, we observed that more than 80% of mB-β₂AR is in the G protein-dependent high-affinity agonist binding state.

Fig. 2 B and C show the emission spectra of mB-β₂AR after reconstitution with Gs. The λ_MAX of mB-β₂AR-Gs fluorescence is similar to that observed for mB-β₂AR following the addition of isoproterenol (Fig. 2A). The effect of Gs on mB-β₂AR fluorescence can be revealed by uncoupling the receptor from Gs by using either GTPγS or GDP. The addition of GTPγS or GDP results in an increase in fluorescence intensity and a 15 nm decrease in the λ_MAX (Figs. 2 B and C and S1 C and D) to a value similar to that observed in unliganded mB-β₂AR before reconstitution (Fig. 2A). GTPγS is more potent at uncoupling mB-β₂AR from Gs compared with GDP, consistent with previous studies indicating that Gs coupled to the β₂AR has a higher affinity for GTPγS than for GDP (24). The concentration-dependent effects of GTPγS and GDP on intensity and λ_MAX of mB-β₂AR are shown in Fig. S1 C and D. The effect of GDP on mB-β₂AR-Gs fluorescence may be unexpected given the prevailing view that the receptor binds preferentially to Gs-GDP and induces GDP release. If mB-β₂AR and Gs-GDP form a complex, this complex does not exhibit the properties of the active state complex as defined above, because cytosolic concentrations of GDP can disrupt Gs dependent high-affinity agonist binding (24). An alternate explanation for these observations is that the β₂AR binds to and stabilizes a small fraction of nucleotide-free Gs that is in equilibrium with Gs-GDP.

Fig. 2E shows a time course of the effect of 200 nM GTP, GTPγS, and GDP on mB-β₂AR fluorescence intensity measured at 450 nm. As expected, all guanine nucleotides induce a rapid increase in fluorescence consistent with the disruption of the active-state mB-β₂AR-Gs complex. The effect at this concentration is larger for GTP and GTPγS compared with GDP. After the initial increase in fluorescence, we observe a decrease in fluorescence for GTP, but not for GDP or GTPγS. This decrease can be explained by the ability of Gs to hydrolyze GTP, but not GDP or GTPγS. Thus, on the initial dissociation of the mB-β₂AR-Gs complex, GTP is hydrolyzed to GDP and some of the mG-β₂AR can reassociate with Gs.

These fluorescence studies show that mB-β₂AR and Gs form an active-state complex (as defined in the introductory paragraphs) in the absence of an agonist. The agonist ISO and Gs induce similar changes in the fluorescence intensity and λ_MAX of mB-β₂AR. These changes are consistent with movement of the fluorophore to a more polar environment that could be achieved by a clockwise rotation and/or outward movement of TM6 relative to TM3 and TM5 (Fig. 1D). This movement is in agreement with the changes observed in rhodopsin by double electron–electron resonance (DEER) spectroscopy (36), as well as conformational differences between rhodopsin and opsin (15), and previous biophysical and mutagenesis studies on the β₂AR (37–39). The fact that both Gs and the agonist ISO induce similar changes in fluorescence is compatible with the hypothesis that they independently induce similar changes in receptor structure and is consistent with the allosteric effect of Gs on agonist affinity and of agonists on Gs activation. Nevertheless, we cannot exclude the possibility that similar changes in mB-β₂AR fluorescence could result from different conformational changes in the β₂AR.

The Effect of Ligand Efficacy on mB-β₂AR-Gs Coupling.

The efficiency with which mB-β₂AR couples to Gs in rHDL particles in the absence of agonist reflects the receptor's intrinsic basal activity. The high effective concentration of Gs and the capacity to trap mB-β₂AR in this active Gs-coupled state in the absence of guanine nucleotides has allowed us to visualize this complex. In cells, however, GDP concentrations exceed 10 μM (28, 29), levels that would destabilize this complex as shown by the capacity of GDP to reverse the effects of Gs on mB-β₂AR (Fig. 2B). Nevertheless, our ability to trap this complex and monitor it through a conformationally sensitive fluorescent reporter allows us to examine the effects of ligand efficacy on the formation and stability of this otherwise transient complex. For the purpose of our discussion, we will assume that a decrease in intensity and increase in λ_MAX of mB-β₂AR-Gs is a reflection of the stability of the active-state complex. This assumption is based on the observation that conditions known to disrupt interactions between β₂AR and Gs such as GTPγS and detergents result in an increase in intensity and a decrease in λ_MAX (Fig. 2B). However, we acknowledge that changes in λ_MAX and intensity may not reflect a proportional change in the affinity of β₂AR for Gs; moreover, it is possible that β₂AR and Gs remain associated in a manner that is not detected by bimane on C265.

Agonist.

In Fig. 3 we compare the effects of an agonist, a neutral antagonist, and an inverse agonist on mB-β₂AR-Gs. Because there is slight variability in receptor to G protein stoichiometry between reconstitutions, these comparisons were performed on the same preparation. Fig. 3A shows the baseline spectrum of the mB-β₂AR-Gs complex before (black) and after (gray) the addition of 10 μM GTPγS, which completely dissociates the complex. The reconstitutions are typically contaminated with a low concentration of GDP needed to stabilize purified Gs. This residual GDP can be removed by the addition of apyrase, a nonselective nucleotide pyrophosphatase. As shown in Fig. 3A, a 40-min treatment with apyrase reverses the uncoupling effect of residual GDP resulting in a small decrease in the intensity of mB-β₂AR (blue spectrum). The addition of the agonist ISO after apyrase treatment causes a further decrease in intensity and a rightward shift in λ_MAX (Fig. 3A, green spectrum). The fact that ISO induces a change in fluorescence intensity on top of that induced by Gs and apyrase suggests that a higher fraction of mB-β₂AR couples to Gs in the presence of agonist and/or the conformation of the β₂AR in the presence of agonist and Gs is different from that in the presence of Gs alone. The maximal effect of ISO on mB-β₂AR-Gs fluorescence occurs at a lower agonist concentration than observed in the absence of Gs (Fig. S2) and is consistent with the effect of Gs on ISO binding affinity (Fig. 2D).

Fig. 3.

The effect of ligands having different efficacies on the stability of mB-β₂AR-Gs. Emission scans were performed on mB-β₂AR-Gs complex obtained from a single reconstitution. For each treatment (apyrase, ISO, alprenolol, ICI, or GTPγS), a baseline scan was performed on mB-β₂AR-Gs alone (black spectrum), and subsequent scans were normalized to this baseline scan. (A) The gray spectrum was taken 10 min after the addition of 10 μM GTPγS to mB-β₂AR-Gs. This treatment uncouples Gs from mB-β₂AR. The blue spectrum shows the effect of a 40-min incubation of mB-β₂AR-Gs with apyrase to remove residual GDP. (A–C) The effect of ligands was determined after 40-min pretreatment with apyrase (blue) followed by a 60-min incubation with the ligand. A: ISO, green; B: alprenolol, purple; C: ICI, red. This experiment is representative of 3 independent experiments.

Neutral antagonist.

By definition, neutral antagonists inhibit binding of agonists, partial agonists, and inverse agonists at the orthosteric binding site of GPCRs, but do not alter their basal receptor activity. It has been difficult to identify true neutral antagonists for the β₂AR, as most compounds display some partial agonist or inverse agonist activity when applied in sensitive signaling assays (40, 41). Nevertheless, alprenolol comes very close to exhibiting the properties of a neutral antagonist, having only very weak partial agonist activity or weak inverse agonist activity, depending on the assay (40, 41). Alprenolol has no significant effect on the fluorescence of mB-β₂AR alone (see below). The effect of alprenolol on mB-β₂AR-Gs was examined following the removal of residual GDP with apyrase. As shown in Fig. 3B, alprenolol (purple spectrum) induces a small decrease in the intensity and red shift in the λ_MAX of mB-β₂AR-Gs that has been treated with apyrase. Thus, alprenolol does not induce dissociation of the mB-β₂AR-Gs complex, and may enhance mB-β₂AR-Gs coupling to a small extent, perhaps related to a low partial agonist activity (41).

Inverse agonist.

In contrast to neutral antagonists, inverse agonists inhibit basal, agonist-independent activation of G proteins. Although the precise mechanism for the action of inverse agonists is not completely understood, they may prevent receptor-G protein complex formation, destabilize preformed complexes, or both. Here, we studied the effect of ICI-118,551 (ICI), one of the most efficacious inverse agonists for the β₂AR (41), on mB-β₂AR-Gs fluorescence (Fig. 3C, red spectrum). mB-β₂AR-Gs was first treated with apyrase to remove residual GDP. Following a 60-min incubation at room temperature, ICI induced only a small reversal of fluorescence change induced by Gs and apyrase (compare blue and red spectra). These data show that ICI is much less effective at disrupting preformed receptor-G protein complexes than is GTPγS (Fig. 3A).

The effect of ligands on β₂AR-Gs complex formation.

The relatively subtle effect of ICI on the stability of the receptor-G protein complex does not explain the inhibitory effect of ICI observed in signaling assays. An alternative mechanism for ICI efficacy, is that the ICI-bound β₂AR does not couple efficiently to Gs. To monitor the effect of ICI on the association of β₂AR and Gs we incubated mB-β₂AR reconstituted in rHDL in the presence or absence of ligands. We then added purified Gs and monitored changes in fluorescence. It should be noted that the amount of Gs that can be added is limited by the detergent in the Gs preparation, consequently the Gs-induced changes in mB-β₂AR are not as large as those observed in Figs. 2 and 3, where the reconstitution was done in the presence of a detergent binding resin. In the absence of ligand, Gs coupling to the β₂AR results in an 18% decrease in intensity and a 5-nm red shift in λ_MAX (Fig. 4A). The addition of 1 μM ISO to mB-β₂AR resulted in a 25% decrease in intensity and a 7-nm red shift in λ_MAX. The addition of Gs resulted in a further decrease in both intensity and λ_MAX (Fig. 4B). In contrast, no change in fluorescence was observed following the addition of Gs to mB-β₂AR that had been incubated for 15 min with 1 μM ICI (Fig. 4C). Like ICI, the neutral antagonist alprenolol had no effect on mB-β₂AR fluorescence. However, in contrast to ICI, alprenolol did not prevent Gs-induced changes in mB-β₂AR intensity and λ_MAX (Fig. 4D). In fact, these changes were larger than those observed for unliganded mB-β₂AR, suggesting alprenolol has partial agonist activity. We also examined carazolol, the partial inverse agonist used to obtain the crystal structures of the β₂AR. We observed a small (approximately 3%) decrease in intensity, but no change in λ_MAX following incubation with 1 μM carazolol (Fig. 4E). Gs induced a further 5% decrease in intensity and a 2-nm shift in λ_MAX. In comparison to unliganded receptor (Fig. 4A), carazolol appeared to partially inhibit Gs-induced changes; however, it was not as effective as ICI. These results show that the inverse agonist ICI is most efficacious in preventing Gs-induced changes in mB-β₂AR, underlying its effectiveness as an inverse agonist in signaling assays.

Fig. 4.

The effect of ligand efficacy on Gs-induced changes in mB-β₂AR fluorescence. Initial emission scans (black spectra) of mB-β₂AR were obtained, then scans were repeated after a 15-min incubation with the indicated ligands (blue spectra). A concentrated solution of Gs was added (1:100 dilution of an 8 mg/mL solution), and emission scans were repeated after 8 min (red spectra). Preliminary studies showed that the effect of Gs was complete at 6 min. Ligands: A, no ligand; B, isoproterenol (Iso); C, ICI; D, alprenolol (Alp); E, carazolol (Cz). These scans are representative of 3 independent experiments.

It should be noted that the lack of changes in mB-β₂AR fluorescence on binding to ICI does not necessarily mean that ICI does not induce or stabilize a conformation distinct from unliganded receptor. It only suggests that conformational changes occurring on ICI binding may not result in a change in the environment around bimane on C265. Previous studies using different fluorescent approaches observed conformational changes on binding ICI to the β₂AR (42) and yohimbine for the α_2A adrenergic receptor (43). The fact that ICI prevents Gs-induced conformational changes observed in the absence of ligand is consistent with a distinct conformation.

Structural Insights into the Active State Complex, Basal Activity, and Ligand Efficacy.

The mechanism by which Gs couples to agonist-free receptor is not known. However, the recent structure of a complex between opsin and the carboxyl-terminal peptide of transducin, suggests that substantial conformational changes are needed in the inactive structure of the β₂AR to permit a similar docking of the carboxyl terminus of Gα_S (Fig. 1 C and D). Evidence from fluorescence lifetime studies on purified β₂AR shows that the unliganded receptor is structurally dynamic (44) and exists in an ensemble of basal conformational states (Fig. 5). Although an active state represents a minor fraction of this ensemble (Fig. 5A), this state would be stabilized by binding to either the agonist ISO (Fig. 5B) or the G protein Gs (Fig. 5C).

Fig. 5.

Conceptual model depicting the dynamic behavior of β₂ARs. In the absence of a ligand, the β₂AR exists in an ensemble of basal states in dynamic equilibrium (blue background). Agonists and inverse agonists bind to and stabilize distinct substates. The nucleotide free form of the G protein Gs can also bind to and stabilize an active state of the β₂AR (green background). The different equilibrium processes between the receptor and its ligands and the receptor and the G protein Gs are displayed (A–H) and are described in detail in the text. Note that for each of these equilibria, the relative size of the arrows indicates the displacement of the reaction.

As shown in Fig. 3A, the agonist ISO (green spectrum) induces a further decrease in the fluorescence intensity and increase in the λ_MAX of the mB-β₂AR-Gs complex. This result could be because of a change in the equilibrium favoring the formation of the complex (Fig. 5D), or to a further change in the structure of the mB-β₂AR-Gs complex. We cannot distinguish between these two possibilities; nevertheless; these observations have implications for efforts to obtain a high-resolution structure of the active state of the β₂AR and possibly other GPCRs. Saturating concentrations of an agonist alone (Fig. 5B) cannot induce the same change in intensity and λ_MAX as that stabilized by both Gs and ISO (Fig. 5D). The results are in agreement with earlier fluorescence lifetime studies on the β₂AR showing that the agonist ISO does not stabilize a single active conformation (44).

Analogous to agonists, the inverse agonist ICI binds to and stabilizes a minor fraction of the ensemble of basal conformational states, in this case corresponding to an inactive conformation (Fig. 5E). The capacity of ICI to prevent formation of the mB-β₂AR-Gs complex (Fig. 4C) suggests that the conformation stabilized by the inverse agonist cannot couple to Gs (Fig. 5F). Similarly, the fact that 1 μM ICI (100-fold greater than the K_d) has little effect on the preformed mB-β₂AR-Gs complex (Fig. 3C) suggests that it cannot bind to Gs-coupled β₂AR (Fig. 5G). This result is in agreement with predictions of the extended ternary complex model (45, 46) and previous studies showing that binding sites for [³H]-ICI-118,551 were reduced in cells expressing high levels of Gs (47). These effects can also be rationalized in light of the crystal structure of the inactive, inverse agonist-bound β₂AR. ISO is 40% smaller in volume than the inverse agonist carazolol present in the β₂AR crystal structure. Thus, the binding site of the ISO-bound receptor will have to readjust to satisfy all of the binding interactions predicted for the agonist ISO (18).

Relevance to Cellular Signal Transduction.

We were able to characterize the properties of the mB-β₂AR-Gs complex by reconstituting purified β₂AR and Gs under conditions that would not be found in a living cell. Reconstitutions were performed at low GDP concentrations, and GDP was further reduced by using apyrase. In the presence of apyrase, virtually all Gs will be in the nucleotide-free state (Fig. 5H). Thus, it is possible to trap both unliganded and agonist-bound mB-β₂AR-Gs complexes. Under these conditions, the inverse agonist is unable to disrupt the complex (Fig. 5G), most likely because of allosteric effects of the G protein on the ligand binding pocket. However, in the context of a cell, where concentrations of GDP may exceed 10 μM, the formation of an active state β₂AR-Gs complex will be a relatively rare event governed by 2 equilibria: 1 for the conformational transition to an active state of the receptor (Fig. 5A), and 1 for the formation of nucleotide free Gs (Fig. 5H). Once the nucleotide-free β₂AR-Gs complex is formed, it would be rapidly disrupted on binding of either GDP or GTP. Thus, although ICI may not be able to disrupt the β₂AR-Gs complex, these complexes are relatively rare and short-lived. ICI efficacy is due primarily to binding to the receptor and stabilizing a conformation that is unable to couple to Gs (Fig. 5E). In contrast, agonists facilitate the formation of the complex by increasing the fraction of β₂AR in an active conformation (Fig. 5B), and possibly by stabilizing a conformation in the β₂AR that allosterically reduces the affinity of Gs for GDP.

In conclusion, our studies examine the allosteric interaction between ligand binding and Gs coupling by using a conformational reporter on the β₂AR, and they provide a structural framework for understanding the concept of basal activity and ligand efficacy. Agonists and Gs induce similar changes in the fluorescence intensity and the λ_MAX of mB-β₂AR-Gs, suggesting they may induce a similar conformational change involving TM6. The complex formed between the β₂AR and Gs in the absence of agonist is stable in the absence of guanine nucleotides. A neutral antagonist has little effect on the formation of the β₂AR Gs complex, whereas an inverse agonist prevents complex formation. Neither antagonist nor inverse agonist promotes dissociation, whereas both GDP and GTP rapidly reverse Gs-induced changes in mB-β₂AR fluorescence. These findings provide insights into the structural basis of drug efficacy, that is, how different chemical structures are ultimately translated into divergent behaviors through the modulation of the interaction between the receptor and the G protein.

Materials and Methods

Expression, Purification, and Labeling of β₂AR.

Construction and characterization of a modified β₂AR where 4 reactive cysteines were mutated (C77V, C327S, C378A, and C406A) was previously described (35). The modified β₂AR was expressed in Sf9 insect cells by using recombinant baculovirus and purified by sequential M1 antibody affinity and alprenolol affinity chromatography as previously described (35). Purified receptor and monobromobimane were mixed at same molarity and incubated overnight on ice in the dark. The fluorophore-labeled receptor was purified right before use by gel filtration.

In Vitro Reconstitution of β₂AR and Gs into rHDL.

Wild-type human apoA-I was purified from human serum as described in detail in Whorton et al. (31). Gs heterotrimer (Gαs_s, his6-β₁, γ₂) was expressed in Sf9 cells and purified essentially as previously described (48). Monobromobimane labeled β₂AR was reconstituted into rHDL particles together with purified Gs heterotrimer as previously described (31). Competition radioligand binding experiments were performed as previously described (31).

Fluorescence Spectroscopy.

Fluorescence spectroscopy experiments were performed on a Spex FluoroMax-3 spectrofluorometer (Jobin Yvon Inc.) with photon counting mode by using an excitation and emission bandpass of 4 nm. For each scan, the final concentration of receptor range from 50–100 nM. For emission scans, excitation was set at 370 nm and emission was measured from 435–485 nm with an integration time of 0.5 s/nm. To determine the effect of ligands, the spectra were taken after 15 min incubation with the drugs. For time course experiments, excitation was at 370 nm, and emission was monitored at 450 nm. All experiments were performed at 25 °C, and the sample underwent constant stirring. Fluorescence intensity was corrected for dilution by ligands in all experiments and normalized to the initial value. Fluorescence intensity was corrected for background fluorescence from buffer and ligands in all experiments. All of the compounds tested had an absorbance of <0.01 at wavelengths between 370 nm and 485 nm at the concentrations used, excluding any inner filter effect in the fluorescence experiments.