Introduction
Discerning the basic mechanisms by which agonists and inverse agonists exert their distinct effects on receptor function is fundamental to understanding signal transduction mediated by GPCRs. To examine whether binding of inverse agonists to GPCRs causes conformational changes resulting in inverse agonism6, 7, 8 (that is, inverse agonist effects), we used a recently established fluorescence resonance energy transfer (FRET) approach to monitor the activation of GPCRs directly in living cells5. This approach relies on the capacity of a GPCR sensor based on cyan and yellow fluorescent proteins, GPCRCFP/YFP, to report in real time, through a fast decrease in FRET, the intramolecular conformational rearrangements associated with receptor activation (see Supplementary Fig. 1 online). We applied this strategy to the 
2AAR, a receptor that couples to the G proteins Gi and Go and binds to several well-characterized inverse agonists9.
We applied saturating concentrations of norepinephrine (agonist) or yohimbine (inverse agonist) sequentially to a single cultured human embryonic kidney (HEK293) cell expressing 2AARCFP/YFP and recorded FRET signals over time (Fig. 1a). Norepinephrine induced a fast decline of the FRET signal, whereas yohimbine increased the FRET signal. This opposite change in the signal suggests that in response to yohimbine the receptor undergoes a conformational change distinct from that seen in response to norepinephrine. Similar signals obtained with three other distinct inverse agonists, rauwolscine, RX821002 and MK-912 (partial response) (Figs. 1b and 2a), confirmed that the increase in the FRET signal represented a general response to inverse agonists.
Figure 1: Conformational changes of the 
2AAR in response to full and inverse agonists.
![Figure 1 : Conformational changes of the |[alpha]|2AAR in response to full and inverse agonists.](/nchembio/journal/v1/n1/thumbs/nchembio705-F1.gif)
(a) Emission intensities of YFP (yellow fluorescent protein) and CFP (cyan fluorescent protein) were recorded simultaneously from a single HEK293 cell expressing 2AARCFP/YFP, and with FRET calculated as the ratio of emission intensities FYFP/FCFP (red). Shown are the changes induced by rapid superfusion with norepinephrine (NE; 100 
M) or yohimbine (Yoh; 300 M). Traces are representative of at least ten separate experiments. (b) Percentage change in the FRET ratio (left) and anisotropy values (right) of 2AARCFP/YFP in intact HEK293 cells measured in the absence (-) or presence of NE (100 M) or rauwolscine (Rau; 300 M). Data indicate the mean 
s.e.m. of four separate experiments.
Figure 2: Action of inverse agonists on 2AARCFP/YFP.
![Figure 2 : Action of inverse agonists on |[alpha]|2AARCFP/YFP.](/nchembio/journal/v1/n1/thumbs/nchembio705-F2.gif)
(a) FRET signals mediated by the inverse agonists yohimbine (Yoh; 100 M), rauwolscine (Rau; 100 M) and RX821002 (RX; 100 M) and by the agonist norepinephrine (NE; 100 M). (b) Effect of the antagonist phentolamine (Phe; 100 M) or the inverse agonist Yoh (100 M) on the FRET signal caused by NE. (c) Relationship between the apparent rate constant kobs and NE (
) or Yoh (
) concentrations. kobs values obtained from fitting the kinetic data of experiments like those of Figure 1a with a monoexponential equation. Data indicate the mean s.e.m. of ten experiments for NE and at least three for Yoh. Results for NE values comparison originate from our previous study5. (d) Plot of kobs values for diverse types of ligands versus the percentage changes they induce in the ratio FYFP/FCFP. Data obtained at saturating concentrations of full agonists: NE (n = 10), UK-14,304 (UK; n = 4); partial agonists: moxonidine (Mox, n = 8), dopamine (DA, n = 8), oxymetazoline (Oxy, n = 4), clonidine (Clo, n = 5); inverse agonists: MK-912 (MK; n = 3), Rau (n = 4), Yoh (n = 7) and RX (n = 4). Note that data for Rau, Yoh and RX were similar, and only data for Yoh are represented.
We further tested whether yohimbine could reverse the receptor's conformational change mediated by norepinephrine. After continuous exposure of a cell to norepinephrine, the addition of yohimbine reversed the FRET signal change generated by norepinephrine and resulted in a signal similar to that caused by yohimbine in the absence of norepinephrine (Fig. 2b). This contrasts with the effect of an antagonist such as phentolamine, which also prevents the signal mediated by norepinephrine (Fig. 2b) but does not cause a conformational rearrangement per se5. Therefore, yohimbine and phentolamine both suppressed the signal produced by norepinephrine, in accordance with their competitive binding nature, but the inverse agonist differentiated itself from the antagonist by causing a conformational rearrangement of the receptor rather than just blocking the binding site.
To clarify whether the FRET changes arose from changes in the orientation of the polarization of the fluorophores or changes in distance, we simultaneously measured in 2AARCFP/YFP fluorophore anisotropies and FRET ratios in response to agonist (norepinephrine) or inverse agonist (rauwolscine) with a microscope setup as described in a published study10 (Fig. 1b). The anisotropy values for CFP and YFP agreed well with those previously established for pure CFP11 and YFP12 in aqueous solution, and they remained unchanged in response to norepinephrine or rauwolscine (Fig. 1b, right), whereas the FRET signal changed in opposite directions (Fig. 1b, left). Thus, the fluorophores showed similar dipole-dipole orientations when the receptor responded to agonist or inverse agonist, suggesting that changes in FRET reflected changes in the distances between the fluorophore moieties. Given the position of the fluorophores in the receptor, we assume that the decrease in FRET in response to agonist reflects a distance separation between the third intracellular loop and the C terminus, whereas the increase in FRET in response to the inverse agonist is consistent with the view that these two receptors' domains come closer together.
The temporal resolution of FRET with increasing ligand concentrations showed that yohimbine-induced conformational switching of the receptor follows a hyperbolic saturation of the rate constant (Fig. 2c). This behavior is consistent with a simple process whereby at low ligand concentration the rate constant (kobs) increased linearly with ligand concentration, indicating that ligand binding was the rate-limiting step. At high ligand concentrations the rate constants approached a constant value, suggesting that a step other than the association of ligand and receptor became rate limiting. This step is consistent with a ligand-mediated conformational change of the receptor. We observed a similar hyperbolic dependence on norepinephrine concentrations in earlier work5, indicating a similar model but for a different conformational change (Fig. 2c). However, the conformational switch responded 
35 times slower to yohimbine than to norepinephrine (Fig. 2c). Thus, not only does the switch induced by the inverse agonist produce a signal opposite to the agonist-induced signal, but the vastly different kinetics suggest that the inverse agonist-induced switch involves a specific conformational change relating to a distinct kinetic pathway.
In the next experiment we tested whether the slow kinetics of receptor conformational changes were characteristic of inverse agonists. The kinetic comparison between structurally distinct low- and high-affinity full agonists (norepinephrine and UK-14,304), partial agonists (dopamine, moxonidine, clonidine and oxymetazoline) and inverse agonists (rauwolscine, yohimbine, RX821002 and MK-912) (see Supplementary Fig. 2 online for chemical structures) suggests that ligands with different efficacies induce a range of conformational changes in the receptor with distinct kinetics (Fig. 2d and Supplementary Fig. 3 online). These different kinetics contrast with structural differences between ligands and with their affinities but are in good agreement with ligand efficacies.
In contrast to other GPCRs that show constitutive activity, we could not detect basal activity of either the wild-type 2AAR or the 2AARCFP/YFP in HEK293 cells at the level of cAMP or G protein inwardly-rectifying K+ channel (GIRK) current (data not shown). This impeded our ability to link structural rearrangements caused by inverse agonists to signal suppression. Therefore, to facilitate the measurement of suppression of basal receptor activities by inverse agonists, we took advantage of a well-characterized threonine-to-lysine point mutation, T373K, in the third intracellular loop adjacent to helix 6 of the 2AAR, which generates a constitutively active mutant (CAM) receptor, 2AARCAM (ref. 13). The 2AARCFP/YFP was thus converted into a CAM receptor, 2AARCFP/YFPCAM. We demonstrated the constitutive activity of 2AARCFP/YFPCAM by three distinct approaches. First, radioligand binding experiments showed that the 2AARCFP/YFPCAM. had an increased affinity for the agonist norepinephrine and a slightly lower affinity for inverse agonists such as yohimbine (Supplementary Fig. 4 online) or rauwolscine (data not shown). These data are in agreement with similar studies done with the wild-type 2AAR and 2AARCAM (ref. 9). Second, cAMP synthesis was significantly suppressed in forskolin-stimulated cells expressing 2AARCFP/YFPCAM (Fig. 3a). Third, coexpression of the parathyroid hormone (PTH) receptor (a Gs-coupled receptor) and the 2AARCFP/YFPCAM reduced the ability of PTH to mediate cAMP accumulation even in the absence of the 2AAR agonist UK-14,304 (Supplementary Fig. 4 online). These data emphasize that 2AARCFP/YFPCAM is constitutively active.
Figure 3: Characterization of the constitutively active receptor.
(a) cAMP measurement of cells expressing 2AARCFP/YFP (wild type (WT), 1 pmol mg-1) or 2AARCFP/YFPCAM (CAM, 0.9 pmol mg-1) and incubated with 10 M forskolin (FSK). FSK-stimulated cAMP in mock-transfected cells was set as 100%, and bars represent the mean s.e.m. of four independent experiments. (b) Energy transfer efficiency (E.T.) values for 2AARCFP/YFP (WT, n = 9) and 2AARCFP/YFPCAM (CAM, n = 6). Significant differences from WT at P < 0.01 (**) by t-test indicated. (c) Comparing switches of 2AARCFP/YFP (dotted trace) and 2AARCFP/YFPCAM (solid trace) in response to norepinephrine (NE; 100 M) or yohimbine (Yoh; 100 M). The traces are representative of three separate experiments. (d) FRET signals of 2AARCFP/YFPCAM for diverse concentration of Yoh. (e) Concentration-response relation for the change in FRET (from data similar to those in Figure 3d, 
) and FSK-stimulated cAMP () in cells expressing 2AARCFP/YFPCAM. Data indicate the mean s.e.m. of three separate experiments.
Photodestruction (that is, photobleaching) of YFP leads to recovery of the CFP emission that corresponds to the energy transfer (Supplementary Fig. 1 online), and this made it possible to determine and to compare the energy transfer efficiencies (and structural consequences) of 2AARCFP/YFP and 2AARCFP/YFPCAM. The mutant receptor showed a lower efficiency of energy transfer (38%) than did 2AARCFP/YFP (45%) (Fig. 3b). This decrease in energy transfer efficiency agrees with a decrease in the level of forskolin-stimulated cAMP in 2AARCFP/YFPCAM –expressing cells (Fig. 3a). The single substitution T373K in the receptor triggers a structural rearrangement that leads to an alteration in the distance between the two fluorophores, apparently resulting in a lower FRET efficiency. These data provide compelling evidence that 2AARCFP/YFP and 2AARCFP/YFPCAM have distinct conformations and that constitutive activation of 2AARCFP/YFPCAM is tightly coupled to the conformational difference observed between the receptor and its constitutively active variant.
Antagonists such as phentolamine or idazoxan induced only insignificant changes in the FRET signal of 2AARCFP/YFP or 2AARCFP/YFPCAM (data not shown). In contrast, in response to saturating concentrations of norepinephrine or yohimbine, the constitutively active receptor showed a pattern of FRET changes similar to those observed for 2AARCFP/YFP but of different amplitude: in 2AARCFP/YFPCAM, changes in FRET were smaller for norepinephrine but higher for yohimbine (Fig. 3c). These differences are understandable if we assume that in the constitutively active receptor much of the conformational activation occurs before agonist binding. Then, the receptor switches from an intermediate active to a fully active state, giving rise to a smaller conformational change.
To verify that structural rearrangements upon inverse agonist binding contribute to inverse agonism, we compared the effects of different concentrations of yohimbine on forskolin-stimulated cAMP in cells expressing 2AARCFP/YFPCAM with changes in FRET. As the extent of FRET increased, the constitutive inhibition of cAMP synthesis was suppressed (Fig. 3d,e). The notable parallel effect between the FRET signal magnitude and changes in cAMP signaling strengthened the conclusion that inverse agonism depends on the receptor's structural rearrangement mediated by inverse agonists.
In summary, inverse agonists signaled a conformational rearrangement in the receptor that accompanied inverse agonism. The differences in the kinetics and character of the conformational changes induced by full agonists and inverse agonists, and also by partial agonists (see ref. 5 and Supplementary Fig. 3 online), imply that these different classes of drugs use distinct molecular switches in the receptor. This demonstrates that the receptors do not function merely as simple 'on-off' switches but rather have several distinct conformational states and can be switched into these states with distinct kinetics.
Methods
Molecular biology and cell culture.
Constructions of the 2A-adrenergic receptor sensor 2AARCFP/YFP and its CAM 2AARCFP/YFPCAM were carried out by established PCR strategies14. Receptor cDNAs were cloned into pCDNA3 (Invitrogen) for transient and stable expression in mammalian cells. HEK293 cells served as the expression systems, and the procedure for the selection of stable cell line has been described15.
Pharmacology.
Ligand binding, receptor number determination and measurement of cAMP were measured as described15, 16. Saturation and competition binding studies were analyzed with Prism 4.0 (GraphPad).
Microscopic FRET measurements.
FRET experiments were done as described5. In brief, cells grown on coverslips were maintained in HEPES buffer (137 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 20 mM HEPES, 0.1% (wt/vol) BSA, pH 7.4) at room temperature and placed on a Zeiss inverted microscope (Axiovert135) equipped with an oil immersion 100
objective and a dual-emission photometric system (Till Photonics). Samples were excited with light from a polychrome IV (Till Photonics). To minimize photobleaching, the illumination time was set to 5–10 ms applied with a frequency between 1 and 75 Hz depending on agonist concentration. FRET was monitored as the emission ratio of YFP to CFP, FYFP/FCFP, where FYFP and FCFP are the emission intensities at 535 15 nm and 480 20 nm (beam splitter dichroic long-pass (DCLP) 505 nm) upon excitation at 436 10 nm (beam splitter DCLP 460 nm). The emission ratio was corrected by the respective spillover of CFP into the 535-nm channel (spillover of YFP into the 480-nm channel was negligible) to give a corrected ratio FYFP/FCFP. FRET between CFP and YFP in cells stably expressing the receptor constructs was also determined by donor recovery after acceptor bleaching. FRET efficiency was calculated according to the equation:
where CFPbefore and CFPafter are CFP emissions before and after photobleaching the YFP by 3–5 min illumination at 500 nm.
Recording of ligand-induced changes in FRET.
To determine ligand-induced changes in FRET, cells were continuously superfused with the HEPES buffer and ligand was applied using a computer-assisted solenoid valve–controlled rapid superfusion device (ALA-VM8 from ALA Scientific Instruments; solution exchange 5–10 ms). Signals detected by avalanche photodiodes were digitalized using an AD converter (Digidata1322A; Axon Instruments) and stored on a PC using Clampex 9.0 software (Axon Instruments). The change in FRET ratio was fitted to the equation r(t) = A (1 - e-t/
), where 
is the time constant (s) and A is the magnitude of the signal. When necessary for calculating , ligand-independent changes in FRET due to photobleaching were subtracted.
Simultaneous measurements of FRET and anisotropy.
Fluorescence anisotropy and FRET were measured with a microscopy setup as described10. The fluorophores were excited with a linear polarized light at 458 nm for the donor and 514 nm for the acceptor. The fluorescence emission was split to its parallel and perpendicular components by a Wollaston prism and into the two emission wavelength ranges for CFP (465–500 nm) and YFP (550–610 nm) by a wedge with dichroic reflective coated surfaces. Intensities were recorded at the plasma membrane, and the full intensity for each fluorophore was calculated as I
+ 2 g I
. The anisotropy was calculated as A = (I - g I)/(I + 2 g I), where g is a correction factor for the relative detection efficiencies between the parallel and the perpendicular channel (g = 
/
) and I and I are the intensities of parallel and perpendicular fluorescence, respectively.
2 adrenergic receptor
