Detection of MOR and _2A-AR heterodimers

We measured the presence of receptor heterocomplexes in transfected human embryonic kidney (HEK) 293 cells by FRET experiments between MOR and _2A-AR C-terminally tagged with yellow fluorescent protein (YFP) and cyan fluorescent protein (CFP), respectively (MOR^YFP and _2A-AR^CFP). The heterodimer formation between MOR^YFP and _2A-AR^CFP caused resonance energy transfer between the fluorescent proteins, of which efficiency was measured by the recovery of the CFP emission after photodestruction (that is, photobleaching) of the YFP (Fig. 1a). Nonspecific FRET due to random distribution and collision between CFP and YFP molecules in the plasma membrane was assessed by expression of pairs of N-terminally membrane-tagged CFP and YFP molecules (CFP^m and YFP^m, respectively). We found 3.5% FRET efficiency between CFP^m and YFP^m (Fig. 1b). At comparable fluorescence levels, we found similar results in cells coexpressing (i) YFP^m and the parathyroid hormone receptor C-terminally tagged with CFP (PTHR^CFP), (ii) _2A-AR^CFP and YFP^m, and (iii) the adenosine A1 receptor C-terminally tagged with YFP (A₁R^YFP) and _2A-AR^CFP (Fig. 1b). This indicates that there is no specific FRET between these molecules. Compared with these control conditions, the FRET efficiency was significantly higher (P < 0.001) for cells expressing _2A-AR^CFP and MOR^YFP, thus indicating the specific presence of hetero-oligomers between MOR and _2A-AR, which is unaffected by the presence of morphine and/or norepinephrine (NE, Compound 2).

Figure 1: Determination of an association between MOR and _2A-AR.

(a) Effect of photobleaching. Emission intensities of YFP (535 nm, black) and CFP (480 nm, gray) recorded from single cells coexpressing MOR^YFP and _2A-AR^CFP using fluorescence microscopy. Emission intensities were recorded before and after YFP (the acceptor fluorophore) was photobleached by 5 min exposure to light at 500 nm. (b) FRET efficiency calculated according to equation (2) from cells expressing a combination of N-terminally tagged CFP and YFP molecules (CFP^m and YFP^m), or C-terminally CFP- or YFP-tagged receptors: CFP^m/YFP^m (n = 8); PTHR^CFP/YFP^m (n = 4); YFP^m/_2A-AR^CFP (n = 4); A₁R^YFP/_2A-AR^CFP (n = 4); MOR^YFP/_2A-AR^CFP (n = 8); ^**P < 0.001. Data indicate the mean s.e.m. (c) Values of FRET efficiency from cells coexpressing MOR^YFP and _2A-AR^CFP, or CFP^m and YFP^m. Data for CFP^m and YFP^m were directly proportional to the YFP emission intensity, whereas FRET efficiency values for MOR^YFP and _2A-AR^CFP followed a hyperbolic function of acceptor intensity.

Full size image (46 KB)

As expected for specific receptor-receptor interactions, we found that the efficiency of energy transfer between MOR^YFP and _2A-AR^CFP, which we measured by donor (CFP) recovery after acceptor (YFP) photobleaching, increases as a hyperbolic function of the concentration of MOR^YFP (as determined by YFP emission intensities; Fig. 1c). Conversely, nonspecific interactions between CFP^m and YFP^m gave a linear increase in FRET efficiency. The saturating behavior is expected for proteins forming specific oligomers^{21, 22} and argues against nonspecific FRET resulting from random collisions due to protein overexpression. These data indicate the formation of hetero-oligomers between MOR and _2A-AR in HEK 293 cells.

Cross-conformational switches between receptors

To determine whether the physical interaction between the MOR and _2A-AR modulates receptor activation, we monitored the formation of the active receptor state using a FRET strategy that we recently developed to directly record conformational changes in GPCRs in real time and in live cells^{17, 19, 20}. As depicted in Figure 2a, this approach relies on intramolecular changes in FRET between CFP fused to the C terminus of the receptor and the membrane-permeant dye FlAsH (fluorescein arsenical hairpin binder, Compound 3) bound selectively to a tetracysteine CCPGCC motif inserted into the third intracellular loop of the receptor^{19, 20} (Supplementary Methods online). Experiments are performed under a fluorescence microscope in which a single HEK 293 cell is selectively excited at 436 nm, and the changes of the CFP and FlAsH emission fluorescences recorded over time allow for analysis of the agonist-induced conformational change of the receptor as it switches from a resting to an active state. As previously reported for various GPCRs^{17, 19, 20}, a decrease in the FRET ratio reflects the response of the activation switch of the receptor to an agonist. We previously showed that the insertion of the FlAsH binding motif into the third loop has no effect on the binding and signaling properties of the receptor²⁰.

Figure 2: Transconformational switching of the _2A-AR by the MOR as a mechanism underlying direct inhibition of receptor activation.

(a) Receptor activation is monitored by recording changes in FRET between CFP (blue circle, the donor) and FlAsH (yellow circle, the acceptor) introduced respectively into the C-terminal tail and the third intracellular loop of the _2A-AR. (b) Time-resolved changes of the FRET ratio F_FlAsH/F_CFP in a single HEK 293 cell expressing the _2A-AR^FlAsH/CFP (left panels) or coexpressing _2A-AR^FlAsH/CFP and MOR (right panels). Yellow and blue traces represent emission intensities of FlAsH and CFP, respectively. The red trace represents the calculated FRET ratio corrected according to equation (1) with the initial value at t = 0 set to 1. Horizontal bars indicate the application of NE (50 M) or morphine (50 M) to the cell. Traces are representative of n 10 experiments. (c) Effect of morphine (50 M) on NE-activated _2A-AR^FlAsH/CFP in cells coexpressing _2A-AR^FlAsH/CFP and MOR and in the presence of pertussis toxin (PTX). The change in the corrected FRET ratio (FRET) was set to 100% (response to 50 M NE) (n = 3). (d) Bars represent the effects of NE and morphine added alone or together on the change in FRET (N_FRET values s.e.m. calculated according to equation (2)) occurring in membranes prepared from HEK cells expressing _2A-AR^FlAsH/CFP and MOR (n = 7). ^**P < 0.05 when comparing the values obtained after addition of NE versus NE + MOR.

Full size image (61 KB)

Addition of NE to a single cell expressing _2A-AR^FlAsH/CFP alone resulted in a fast reduction of the FRET signal by 18 2.8% (n = 20) with a time constant of = 55 5 ms (n = 20) (Fig. 2b and Supplementary Fig. 1a online). This rapid signal presumably reflects a conformational rearrangement that takes place as the receptor switches from an inactive to an active state, as previously shown^{17, 19, 20}. The addition of morphine in the presence of NE had no effect on _2A-AR^FlAsH/CFP activation (Fig. 2b). In cells coexpressing _2A-AR^FlAsH/CFP and MOR, morphine had no effect on the FRET signal in the absence of NE. However, in these cells morphine suppressed the NE-induced FRET signal by 22.9 2.1% (n = 15, P < 0.001; Fig. 2b and Supplementary Fig. 1a). The specific MOR antagonist D-Phe-cyclic[Cys-Tyr-D-Trp-Arg-Thr-Pen]-Thr-NH₂ (CTPA) did not alter the NE-mediated FRET signal, which indicates that the inhibitory effect is agonist dependent.

We obtained similar effects after pretreatment of cells with pertussis toxin in order to inactivate G_i (Fig. 2c and Supplementary Fig. 1b). This indicates that the inhibition of the FRET signal was independent from a receptor–G protein interaction. These data suggest that following morphine binding, the MOR alters the activation of the _2A-AR through a conformational change transmitted from the MOR to the _2A-AR. This transconformational effect occurred reversibly (Supplementary Fig. 1c) and rapidly (Supplementary Fig. 1d), with a minimal time constant of 0.45 0.1 s (n = 5) at a saturating concentration of morphine (Supplementary Fig. 1e). This activation constant is slightly faster than that of G_i activation by _2A-ARs²⁰ and is thus compatible with a conformational change by a protein directly contacting a GPCR.

Next, we confirmed that the transconformational change across receptors originates at the level of the receptors themselves, and is not caused by interactions with other cytosolic proteins such as G proteins, kinases or arrestins. To exclude interactions with these proteins, we studied the _2A-AR in isolated cell membranes treated with 6 M urea and prepared from HEK 293 cells expressing the receptor biosensor alone or with the MOR. Under these conditions, morphine had inhibitory effects on the magnitude of NE-induced FRET change that were similar to the inhibitory effects in intact cells (Fig. 2d). Thus, the transconformational effect is not caused by interactions of receptors with other proteins, but originates in the active conformation of the receptor itself.

Receptors' cross-switches control G protein activation

The fact that morphine regulated the level of _2A-AR activation in cells expressing both the _2A-AR and the MOR, but not in cells expressing _2A-AR alone, suggests functionally relevant interactions between the MOR and the _2A-AR. Given that agonist interaction with the receptor's binding site triggers conformational changes^{17, 19, 23, 24}, we propose that the propagation of these conformational changes alters the structural conformation in the neighboring receptor, which in turn might modify the coupling and activation of the corresponding G protein.

We examined this possibility by recording intermolecular changes using a recently established FRET approach to directly monitor heterotrimeric G protein activation in live cells^{18, 20} (Fig. 3a). We measured changes in FRET between YFP and CFP inserted into the G protein subunits G_i and G₂ (G_i^CFP/YFP), respectively, as a readout of G protein activation. Application of NE (100 nM) to single cells expressing _2A-AR induced a fast decrease in FRET, reflecting conformational and/or dissociational events during G_i activation (Fig. 3b). We obtained signals of similar amplitudes with morphine (100 nM) acting on cells expressing the MOR, which confirms the ability of morphine to activate G_i (Fig. 3b). Morphine did not affect G_i activation via _2A-AR, nor did NE affect G_i activation via MOR when the receptors were expressed alone (Fig. 3b). In contrast, in cells coexpressing the two receptors, morphine inhibited the FRET signal of G_i produced by NE (Fig. 3c), with an average inhibition of 25 2.5% (P < 0.01, n = 12; Supplementary Fig. 2a,c online). Such an inhibition can be observed in the reverse reaction when morphine acts on NE-activated _2A-AR (Fig. 3c), though with a much smaller effect. Indeed, morphine (100 nM) impaired NE-mediated G_i activation by 8.1 3.1% (P < 0.05, n = 8; Supplementary Fig. 2b,c). The absence of an inhibitory effect of adenosine (Compound 4) on NE-mediated G_i activation in cells coexpressing _2A-AR^FlAsH/CFP and the A₁R, another G_i-coupled receptor that does not interact with _2A-AR, firmly supports that the observed effects on G_i activation and receptor activation result from a direct interaction between the MOR and the _2A-AR (Supplementary Fig. 3 online). Thus, this experiment excludes the possibility that the transinhibition effect results from a nonspecific effect of coexpressed GPCRs sequestering G proteins. We then verified that the inhibition of _2A-AR activation correlates with the subsequent inhibition of G_i activation in response to increasing concentrations of morphine (Fig. 3d), thus indicating that the transconformational effect mediated by morphine impairs G_i activation by NE.

Figure 3: Inhibition of G_i protein activation by transconformational switches between MOR and _2A-AR.

(a) For G_i activation, FRET was measured between YFP-labeled G_i and CFP-labeled G₂ subunits as previously described^{18, 20}. (b) Representative FRET experiments showing direct effect of ligands (100 nM) in a single HEK 293 cell coexpressing G_i^CFP/YFP (G_i1^YFP + G₁₂^CFP) and _2A-AR (left panel) or MOR (right panel). G_i activation is associated with a decrease in the FRET signal (red) defined as the FRET ratio of emission intensities of F_YFP/F_CFP and corrected according to equation (1) with the initial value at t = 0 set to 1. Horizontal bars represent the duration of agonist application (representative experiments; n = 5). (c) Experiments similar to those in b showing the effect of NE and morphine added alone or together on HEK cells coexpressing _2A-AR and MOR (representative experiments; n 8). (d) Concentration-response relation for morphine on the change in FRET for receptor activation (black circles, from data similar to those in Figure 2b) and on the change in FRET for G_i activation (red circles, from data similar to those in c) produced by NE in cells expressing _2A-AR^FlAsH/CFP and MOR. Data indicate the mean s.e.m. of five separate experiments.

Full size image (66 KB)

Inhibition of MAP kinase activity by receptor cross-talk

Next, we assessed whether the ability of morphine to inhibit _2A-AR–mediated G_i activation regulates downstream signaling. To this end, we studied responses of NE and morphine on receptor-mediated activation of the mitogen-activated protein (MAP) kinases extracellular signal–regulated kinases 1 and 2 (ERK1/2). Application of either morphine or NE alone to cells coexpressing MOR and _2A-AR stimulated ERK1/2 phosphorylation in a concentration-dependent manner. Half-maximal activation occurred at 21 0.2 nM (n = 3) for NE, and at 3 0.3 nM (n = 4) for morphine (Fig. 4a,b and Supplementary Fig. 4 online). However, the efficacy of NE to mediate ERK1/2 phosphorylation decreased markedly when morphine was co-applied, and this inhibitory effect of morphine is again concentration dependent (Fig. 4c). Morphine did not modulate ERK1/2 activity of control cells expressing _2A-AR alone, which indicates that these effects are dependent on the coactivation of both receptors by their respective agonists. Thus, signaling of one receptor in a MOR–_2A-AR heterodimer is inhibited by the activation of the other receptor, and the simultaneous action of NE and morphine produces a response that is notably different from the expected additive effects.

Figure 4: Concentration-response relation for ligand-mediated phosphorylation of ERK1/2 in cells coexpressing _2A-AR^FlAsH/CFP and MOR.

(a–c) The ligands are NE (a), morphine (b) and morphine with NE (c). The data shown are mean s.e.m. of four separate experiments; ^*P < 0.05. (d) Concentration-response relation for morphine on the FRET signal of G_i^CFP/YFP (gray circles, from data similar to those in Figure 3c) and level of ERK1/2 phosphorylation (open circles, from data similar to those in c) produced by NE in cells expressing _2A-AR and MOR. Data indicate the mean s.e.m. of five experiments for G_i^CFP/YFP and four experiments for ERK1/2 phosphorylation.

Full size image (54 KB)

The concentration-response analyses for NE-mediated G_i activation and stimulation of ERK1/2 phosphorylation in cells expressing _2A-AR and MOR (Supplementary Fig. 5 online) show a close relationship with a very good correlation (r² = 0.98, P = 0.0093) between extents of ERK1/2 phosphorylation and G_i protein activation. This indicates that the changes in the FRET signals of G_i^CFP/YFP reflect the ability of the NE-occupied _2A-AR to stimulate a downstream signal response as measured by ERK1/2 phosphorylation. We observed that the inhibitory effect of different concentrations of morphine on the capacity of NE to stimulate ERK1/2 phosphorylation in coexpressing cells correlates well with changes in FRET of G_i^CFP/YFP and _2A-AR^FlAsH/CFP (r² = 0.98, P < 0.05; Fig. 4d and Supplementary Fig. 6 online). Indeed, as the extent of G_i^CFP/YFP activation measured by FRET decreased, the stimulation of ERK1/2 phosphorylation was suppressed (Fig. 4d). These parallel inhibitory effects suggest that the modulation of _2A-AR–mediated ERK1/2 activity by MOR activation depends directly on a corresponding inhibition of NE-induced G_i activation. We also observed parallel inhibitory effects between FRET signal magnitudes for _2A-AR and G_i, and changes in ERK phosphorylation (Supplementary Fig. 6). This supports the conclusion that inhibitory effects on _2A-AR signaling (G protein activation, ERK phosphorylation) depend on the capacity of morphine to induce structural rearrangements on the _2A-AR by transconformational switches that are transmitted from the MOR to the _2A-AR.

Fluorescent protein constructs and cell culture.

Constructions and functional characterization of _2A-AR^FlAsH/CFP and G_i^CFP/YFP have been reported previously^{18, 19}. Fusion of GFP variants to the C termini of the parathyroid human receptor type 1 (PTHR^YFP and PTHR^CFP) and opioid receptor (MOR^YFP) were performed by DNA recombinant techniques³⁷. Membrane-targeted EYFP (YFP^m) and ECFP (CFP^m) were generated by fusing their N-terminal complementary DNA to cDNA encoding the lipid modification site MGCINSKRKD, as previously reported³⁸. HEK 293 cells served as the expression system, and cells stably expressing _2A-AR and _2A-AR^FlAsH/CFP have been previously described^{19, 20}.

Microscopic FRET measurements.

FRET experiments were performed as previously described^{17, 19}. 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% (w/v) bovine serum albumin (BSA), pH 7.4) at room temperature (22 °C) and placed on a Zeiss inverted microscope (Axiovert 200) equipped with an oil immersion 100 objective and a dual emission photometric system (TILL Photonics). Samples were excited with a xenon lamp from a polychrome IV or V (TILL Photonics). To minimize photobleaching, the illumination time was set to 5–15 ms applied with a frequency between 1 and 75 Hz dependent on agonist concentration. FRET was monitored as a YFP/CFP emission intensity ratio upon excitation at 436 nm (filter 436 10 nm and a beam splitter dichroic long-pass (DCLP) 460 nm). The emission fluorescence intensities were determined at 535 15 nm (YFP) and 480 20 nm (CFP) with a beam splitter DCLP of 505 nm. The FRET ratio for single experiments was corrected according to equation (1):

where F_YFP^ex436/em535 and F_CFP^ex436/em480 represent respectively the emission intensities of YFP (recorded at 535 nm) and CFP (recorded at 480 nm) upon excitation at 436 nm; a and b represent correction factors for the bleed-through of CFP into the 535 nm channel (a = 0.35) and the cross-talk due to the direct YFP excitation by light at 436 nm (b = 0.06). F_YFP^ex500/em535 represents the emission intensity of YFP (recorded at 535 nm) upon direct excitation at 500 nm, and was recorded at the beginning of each experiment. Note that the bleed-through of YFP into the 480 nm channel was negligable. For each measurement, changes in fluorescence emissions due to photobleaching were subtracted. To ensure that CFP- and YFP-labeled molecule expression were similar in examined cells, we performed experiments in cells displaying comparable fluorescence levels.

The means of FRET experiments were calculated according to equation (2) (ref. 39), which normalizes for different expression levels of CFP and YFP molecules:

FRET between CFP and YFP in cells expressing the receptor constructs was also determined by donor recovery after acceptor bleaching. When CFP and YFP exhibited FRET, then photobleaching of YFP by direct illumination at 500 nm increased CFP emission at 480 nm. This was exactly the case when MOR^YFP and _2A-AR^CFP were coexpressed in the same cell (Fig. 1a). The emission intensity of CFP was first recorded at 436-nm excitation (CFP_before), followed by direct illumination of YFP at 500 nm for 5 min. Subsequently, the emission intensity of CFP was recorded again (CFP_after). FRET efficiency was calculated according to Eq. (3), as previously described^{21, 24}:

The application of the photobleaching protocol to cells expressing only CFP at the membrane did not show an increase in the emission intensity of CFP.

Recording of agonist-induced changes in FRET.

To determine agonist-induced changes in FRET, cells were continuously superfused with the HEPES buffer and the agonist was applied using a computer-assisted, solenoid-valve-controlled rapid superfusion device (ALA-VM8, ALA Scientific Instruments; solution exchange 5 to 10 ms). Signals detected by avalanche photodiodes were digitalized using an analog to digital converter (Digidata1322A, Axon Instruments) and stored on PC computer using Clampex 9.0 (Axon Instruments). Data were analyzed using the programs Origin (OriginLab Corp.) and Prism 4.0 (GraphPad).

Statistics.

Data are presented as mean s.e.m. Single group comparisons were analyzed with Student's t-test. Statistical significance was set as P < 0.05. For ERK1/2 assays, statistics were evaluated by ANOVA analysis and Bonferroni post-test.

Accession codes.

Protein Data Bank: The structure of rhodopsin and the structure of inactive heterotrimeric G_i protein were deposited as part of previous studies under PDB codes 1GZM and 1GG2, respectively.

Note: Supplementary information and chemical compound information is available on the Nature Chemical Biology website.

Author contributions

J.-P.V. designed, performed and supervised experiments and wrote the manuscript with support from M.J.L.; V.O.N. performed most of the experiments and analyzed data with J.-P.V.; K.L. performed and analyzed ERK1/2 assays; S.F. and Z.Z. contributed to experiments; all authors discussed the results and commented on the manuscript.