Lifetime of muscarinic receptor–G-protein complexes determines coupling efficiency and G-protein subtype selectivity

Significance

The family of G-protein–coupled receptors (GPCRs) is the largest receptor family, and each member detects specific ligands, which in turn, activate selected members of one or more G-protein families. We developed a broadly applicable assay to study the specificity of GPCR–G-protein interaction in a subtype selective manner. We were able to quantitatively measure both association and dissociation rates of G proteins from agonist-activated GPCRs. We found that the stability of the receptor–G-protein complex is an inherent property of both interaction partners independent on the agonist used. Furthermore, its lifetime correlated closely with the ability of the receptor to activate the corresponding G-protein subtype.

Abstract

G-protein–coupled receptors (GPCRs) are essential for the detection of extracellular stimuli by cells and transfer the encoded information via the activation of functionally distinct subsets of heterotrimeric G proteins into intracellular signals. Despite enormous achievements toward understanding GPCR structures, major aspects of the GPCR–G-protein selectivity mechanism remain unresolved. As this can be attributed to the lack of suitable and broadly applicable assays, we set out to develop a quantitative FRET-based assay to study kinetics and affinities of G protein binding to activated GPCRs in membranes of permeabilized cells in the absence of nucleotides. We measured the association and dissociation kinetics of agonist-induced binding of G_i/o, G_q/11, G_s, and G_12/13 proteins to muscarinic M₁, M₂, and M₃ receptors in the absence of nucleotides between fluorescently labeled G proteins and receptors expressed in mammalian cells. Our results show a strong quantitative correlation between not the on-rates of G-protein–M₃–R interactions but rather the affinities of G_q and G_o proteins to M₃–Rs, their GPCR–G-protein lifetime and their coupling efficiencies determined in intact cells, suggesting that the G-protein subtype-specific affinity to the activated receptor in the absence of nucleotides is, in fact, a major determinant of the coupling efficiency. Our broadly applicable FRET-based assay represents a fast and reliable method to quantify the intrinsic affinity and relative coupling selectivity of GPCRs toward all G-protein subtypes.

G-protein–coupled receptors (GPCRs) are involved in many physiological functions and represent the largest receptor family, accounting for more than several hundred members. These receptors couple to four major G-protein classes (G_i, G_q/11, G_s, and G_12/13) (1), which are defined by the type of Gα subunit. In total, we have a pool of 16 α, 5 β, and 12 γ subunits (2). At the same time, active G proteins are able to interact with more than 25 different effectors, such as adenylyl cyclases, ion channels (3), PLC, RhoGEFs (4, 5), and PDE (4), further enhancing the complexity of GPCR signaling. Although many classes of G proteins exhibit a high degree of selectivity by the given GPCR, it has been shown that the same receptor can couple to more than one class of G proteins (6–11). Such a large number of potential combinations gives rise to the question of how the receptor actually finds the right G protein.

Breakthroughs in GPCR crystallization gave detailed insight into the GPCR structures (12, 13) and GPCRs’ activation and interaction with downstream partners. However, the selectivity of receptor–G-protein coupling and its underlying mechanisms remain unclear (14, 15). In our study, we decided to focus on muscarinic receptors (mAChRs), which belong to class A GPCRs and constitute a family with five subtypes (16). M₁– and M₃–Rs are classic G_q coupling GPCRs (17–20); however, changes in the concentration of second messengers after treatment with pertussis and cholera toxins (17, 21–25) suggest a possible coupling to G_i and G_s proteins. In contrast, M₂–R is a classic G_i family-coupled receptor (19, 23, 26). Despite the fact that mAChRs couple to different G-protein classes, the comparison of crystal structures of these GPCRs revealed only small differences in transmembrane helices (12, 13, 27), meaning that the current knowledge of receptor structures restricts a prediction of the coupling to a particular G-protein class or subtype (28, 29).

In accordance with ternary complex model agonist, receptor and G protein remain stably coupled until GTP binds to the Gα subunit and immediately triggers G-protein activation and its dissociation from the receptor (30, 31). We hypothesized that the mechanism underlying the coupling selectivity must be encoded in the receptor–G-protein interaction. Therefore, we set out to measure the kinetics and affinity of the GPCR–G-protein interaction in the absence of nucleotides to determine the affinity of the G proteins to the receptors in a quantitative manner. Due to the high concentration of nucleotides in the cytosol, the ternary complex has a very short lifetime (32); therefore, previous methods to measure the GPCR–G-protein affinity were mostly based on biochemical assays and required complicated protein purification steps (8, 33–35). In this study, we establish a reliable FRET-based method for quantification of the G-protein affinity to the receptor in a regular plasma membrane environment. With this method, we measured on- and off-rates as well as steady-state binding curves for the agonist-driven interaction of GPCRs with representative G proteins from all four classes.

Results

Studying Ternary Complex Formation by Means of FRET.

To investigate the affinity of receptor–G-protein interaction, the dynamics of G proteins binding to activated GPCRs and their subsequent dissociation were analyzed by FRET under conditions of GTP depletion. HEK293T cells expressing M₃–R–YFP, Gα_q, Gβ₁ subunits, and CFP-Gγ₂ were subjected to single-cell FRET imaging (Fig. 1A), similar to the process described previously (36). In intact non–GTP-depleted cells, we observed a small and rapidly reversible increase in FRET as also shown previously for other receptors (Fig. 1 C and E) (37). We then permeabilized cells by 2-min exposure to 0.05% saponin (Fig. 1 B, D, and F) to deplete membranes from nucleotides, thus allowing the development of relatively stable agonist–receptor–G-protein complexes. The amplitude of FRET signal in permeabilized cells was substantially higher, indicating a largely increased occupancy of receptors with G proteins (Fig. 1F). Most importantly, after withdrawal of agonist, we observed much slower dissociation kinetics of G_q from M₃–R in absence of nucleotides (Fig. 1E vs. Fig. 1F), which reflects the high intrinsic affinity of G proteins to active receptors under nucleotide-depleted conditions. The rapid drop of the FRET signal in response to GTPγS indicates nucleotide-dependent fast dissociation of the remaining receptor–G-protein complexes (Fig. 1F).

Fig. 1.

Assay to measure G-protein–receptor interaction. G_q-protein binding to the M₃–R was measured either in intact (A, C, and E) or permeabilized (B, D, and F) cells by means of brief exposure to saponin as schematically illustrated in B. HEK293T cells transfected with M₃–R C-terminally labeled with YFP, Gα_q-WT, Gβ₁-WT, and Gγ₂ N-terminally labeled with CFP (C–F) were subjected to confocal microscopy and imaged for the CFP fluorescence before (C) and after membrane permeabilization (D). (Scale bars, 10 μm.) For FRET measurements, cells were excited at 430 nm, and YFP and CFP emission was simultaneously imaged using a dual-emission fluorescence microscope. The YFP/CFP emission ratio derived from a single cell was calculated and plotted over time (E and F), and the superfusion of cells with agonists or GTPγS occurred as indicated.

Affinity of G Protein to mAChRs Determines GPCR–G-Protein Selectivity.

Based on reports that M₃–Rs may also couple to G_i/o and even G_s proteins (38), we wanted to quantify the selectivity of M₃–Rs binding to four different classes of G proteins under conditions of nucleotide depletion. Therefore, we compared the agonist [10 µM acetylcholine (ACh)]-evoked FRET signal between fluorescent M₃–R and fluorescent Gβγ subunits when Gα_q, Gα_o, Gα_s, Gα₁₃, or pcDNA3 instead of Gα was transfected (Fig. 2A). As depicted in Fig. 2A, there were no significant differences in the small agonist-evoked FRET signal in cells transfected with Gα_s or Gα₁₃ compared with those that were transfected with empty vector instead of cDNA encoding for Gα subunits, indicating no detectable interactions. Although the binding of Gα_s and Gα₁₃ to M₃–R was not detected, these G proteins did bind with high affinity to receptors, such as β-andrenergic receptors (β₁-AR and β₂-AR) and thromboxane TxA₂ receptor (TP-R) (SI Appendix, Fig. S1), in accordance with the literature data. Correspondingly, we could not detect G₁₃ or G_s activation via stimulation of M₃–R (SI Appendix, Fig. S2). However, we observed a robust agonist-induced FRET signal in cells transfected with Gα_o, which was comparable in amplitude with signals obtained in Gα_q-expressing cells (Fig. 2A, pink vs. black). In experiments studying M₃–R interaction with either G_q or G_o proteins in the presence of high or low concentrations of GDP or GTP, we verified accurate control of nucleotides in permeabilized cells (SI Appendix, Fig. S3) and also verified the absence of nucleotide-sensitive preassociation of receptors and G proteins (SI Appendix, Fig. S4).

Fig. 2.

Selectivity of G protein binding to M₃–R. (A) Agonist-dependent association and dissociation of M₃–R with G_o, G_q, G_s, or G₁₃ proteins were measured after nucleotide depletion by means of FRET similar to that described in Fig. 1F. Average traces of YFP/CFP emission ratio (normalized to initial values) reflecting FRET between YFP-labeled M₃–R and CFP-labeled Gγ₂ subunit are illustrated when Gα_o-WT (pink curve), Gα_q-WT (black), G_s (purple), G₁₃ (red), or empty pcDNA3 (dark green) was cotransfected. A much faster decay of the FRET signal after withdrawal of agonist was detected for G_o proteins compared with G_q. (B) The constants of dissociation (k_off; s⁻¹) for G_o and G_q proteins from M₃–R are shown (magenta, n = 14; black, n = 30, respectively). (C) Observed association kinetics (k_obs; s⁻¹) of G_q protein with M₃–R under GTP-depleted conditions were plotted over different CCh concentrations and fitted by a hyperbolic function. (D) Measured k_on value of M₃–R–G_q association kinetics (light gray, G_q meas.) obtained from fitted hyperbolic function using Eq. 2 compared with ${\text{k}}_{\text{on}}^{\text{C}}$ of G_q (black, G_q calc.) and G_o (magenta, G_o calc.), which were calculated based on steady-state experiments data as shown in Eq. 3. All data are plotted as means ± SEM for each condition; n of each experiment is shown in parentheses if not indicated. Statistical anlysis was performed using one-way ANOVA followed by Student’s t test (***P <0.001).

The comparison of G_q and G_o dissociation rates from M₃–R during the withdrawal of agonist in the absence of nucleotides revealed a significant 13-fold difference (0.009 ± 0.002 and 0.116 ± 0.019 s⁻¹, respectively) (Fig. 2B and Fig. 4A, exponential fit example), which indicates a lower affinity of the M₃–R–G_o complex. Moreover, since we actually measured interaction of fluorescent Gβγ with fluorescent receptors using overexpressed native Gα subunits for a more accurate comparison between receptor–G-protein coupling, we also performed experiments with fluorescently labeled versions of both Gα_q and Gα_o as a control. No labeling-associated differences in the dissociation kinetics for both G proteins were detected (SI Appendix, Fig. S5). Similarly, long lifetimes of receptor–G-protein complexes were observed for other receptors, such as β₁-_, β₂-, and α_2A-ARs as well as TP-R, only for G-protein subtypes, which are known to be activated by the respective receptor (SI Appendix, Fig. S1). Furthermore, we measured the association kinetics of M₃–R with G_o and G_q proteins under nucleotide-depleted conditions in dependence of the agonist concentration (Fig. 2 C and D). Based on generally accepted agonist–receptor occupancy models (39, 40), the rates of receptor–G-protein interaction (k_obs) should increase with an exponential correlation (Eq. 1) to the agonist concentration applied. Taking into account that the exponential function is a subcase of a particular hyperbolic sector, k_on was calculated as a first derivative of k_obs described as hyperbolic function (Eq. 2):

$${\text{k}}_{\text{obs}}=\frac{{\text{P}}_1\times \text{x}}{{\text{P}}_2+\text{x}}\left({\mathrm{s}}^{-1}\right)$$ [1]

$${\text{k}}_{\text{on}}={\left({\text{k}}_{\text{obs}}\right)}^{\prime }=\frac{{\text{P}}_1\times {\text{P}}_2}{{({\text{P}}_2+\text{x})}^2}\hspace{0.17em}\left({\mathrm{s}}^{-1}·{\mathrm{mM}}^{-1}\right)$$ [2]

$${\text{k}}_{\text{on}}^{\text{C}}=\frac{{\text{k}}_{\text{off}}}{{\text{EC}}_{50}}\left({\mathrm{s}}^{-1}·{\mathrm{mM}}^{-1}\right).$$ [3]

Moreover, the k_on values for the G_q, G_o, and G_i2 were also calculated $\left({\text{k}}_{\text{on}}^{\text{C}}\right)$according to Eq. 3 based on the determined k_off values and the apparent EC₅₀ values measured under steady-state conditions in permeabilized cells (Eq. 3, Fig. 3A, and SI Appendix, Fig. S6). Both k_on and ${\text{k}}_{\text{on}}^{\text{C}}$ values for G_q (Fig. 2D) seemed to be relatively similar (22.40 and 14.04 s⁻¹·mM⁻¹, respectively). We also attempted to experimentally determine the on-rate for the M₃–R–G_o interaction (SI Appendix, Fig. S7); however, in this case, the apparent on-kinetics were faster, and if plotted against the agonist concentration, we failed to fit the data to a simple one-component hyperbolic function (SI Appendix, Table S1). Obviously, the measured and calculated association kinetics did not reflect the differences in coupling preference for G_q over G_o proteins. This suggests—at least for the M₃–R—that the lifetimes of the ternary complexes play a major role in the determination of receptor–G-protein selectivity.

Fig. 4.

Comparison of the stability of G-protein–receptor complexes. Decay kinetics of receptor–G-protein complexes were determined in response to agonist withdrawal for M₁–, M₂–, and M₃–Rs and different members of the G_i/o- and G_q-protein family in experiments similar to those shown in Fig. 1F. (A) Examples of exponential curve fittings to the decline in the YFP/CFP ratio after agonist withdrawal for M₁–R and G_i3 (blue) or G_q (black) protein. The constants of dissociation (k_off; s⁻¹) of G_i1, G_i2, G_i3, G_o, and G_q proteins (green, orange, blue, magenta, and black, respectively, in all bar graphs) were calculated for M₁– (B), M₂– (C), and M₃–Rs (D). Calculated k_off (s⁻¹) values, including the replication numbers (n), are given in SI Appendix, Table S2. The data are represented as means ± SEM. (Data for G_o and G_q are taken from Fig. 2B for comparison.) Normalization of alterations in FRET was performed as described in Fig. 2. Statistical analysis was performed using one-way ANOVA followed by Dunnett’s multiple comparison test (**P < 0.01; ***P < 0.001; n.s., P ≥ 0.05).

Fig. 3.

Comparison of receptor–G-protein binding and activation of G proteins. (A) Concentration–response curves of G_q (black) and G_o (pink) binding to M₃–R under GTP-free conditions. FRET imaging in single permeabilized cells was performed as described in Fig. 1F. Increasing concentrations of CCh were applied as shown in a representative trace of a G_q-expressing cell (B). The change of YFP/CFP emission ratio was normalized to the amplitude observed on application of saturating CCh concentration (1 mM; maximum response) and the YFP/CFP ratio measured under the application of GTPγS (minimum response). To achieve steady-state conditions at low CCh concentrations (under 100 nM), cells were incubated for 10 min. (C and D) Concentration–response curves of G_q and G_o activation by M₃–R were determined in intact cells using established G-protein FRET assays (61). (D) Representative traces of FRET between YFP-labeled Gα subunit and CFP-labeled Gγ₂ subunit were measured in duplicates (overflow technique) simultaneously in 12 individual wells derived from a 96-well plate and normalized to the maximal response (1 mM CCh). All concentration–response data are represented as means ± SEM for each condition; n of each experiment is shown in parentheses. Statistical analysis was performed by one-way ANOVA with Bonferroni post hoc test (P < 0.05).

The Difference in G_o and G_q Affinity to M₃–R Reflects Coupling Efficiency.

To experimentally verify whether the measured k_on and k_off were realistic, we next determined steady-state binding curves for the carbachol (CCh)-induced nucleotide-free complex formation of M₃–R with Gα_q- or Gα_o-containing G proteins in dependence of agonist concentration by determining the stable plateau reached after agonist application (Fig. 3 A, black vs. pink and B). The concentration–response curve of G_o binding to M₃–R was found to be 16.5-fold (right shifted in comparison with G_q; EC₅₀ = 10.68 ± 0.32 and 0.64 ± 0.04 µM, respectively), which indicates a higher affinity of binding of G_q to M₃–R compared with G_o. Under these conditions, we determined the G-protein/receptor expression ratio to be two- to sixfold (SI Appendix, Fig. S8) and also detected no major alterations of Gβγ expression in dependence of the coexpressed Gα subtype (SI Appendix, Fig. S9).

To test whether the stability of the ternary complex correlates with the efficiency of G-protein activation, we determined G_o- and G_q-protein coupling efficiency to M₃–R (Fig. 3 C and D) by measuring FRET between YFP-labeled Gα and CFP-labeled Gβγ subunits (41) in thousands of nonpermeabilized adherent transiently transfected cells (Fig. 3D) in a 96-well format. Similar to the binding data, the concentration–response curve of Gα_o activation (EC₅₀ = 0.79 ± 0.01 µM) by M₃–R was 6.6-fold right shifted in comparison with Gα_q activation (EC₅₀ = 0.12 ± 0.01 µM) (Fig. 3C, pink vs. black). Considering our findings regarding the on-kinetics of G_o and G_q binding to activated M₃–R, this result shows that the affinity of the mAChRs–G-protein interaction underlying G-protein selectivity is primarily reflected by the stability (lifetime) of the complex and correlates closely with the coupling efficiency. Moreover, we measured dissociation kinetics of M₃–R–G_i/o proteins complexes in dependence of Gα_i subtypes and found no significant differences for Gα_i1, Gα_i2, and Gα_i3 (Fig. 4D and SI Appendix, Fig. S10A). Similar results were obtained for M₁–R (Fig. 4 A and B and SI Appendix, Fig. S10 B and E).

To test for the G_i/o-subtype selectivity of M₂–R, we repeated similar experiments for M₂–R. Here, we observed a higher amplitude and slightly enhanced complex stability for Gα_o and Gα_i2 compared with Gα_i1 and Gα_i3, as the calculated k_off values of Gα_o and Gα_i2 from M₂–R were twofold smaller in comparison with Gα_i1 and Gα_i3, a result that correlates with the G-protein homology within the G_i family (Fig. 4C and SI Appendix, Fig. S10 D and F). Also, as expected, no binding of Gα_q to M₂–R in the absence of nucleotides was detected (SI Appendix, Fig. S10C). A summary of all k_off values is shown in SI Appendix, Table S2.

Agonist Affinity Is Not Crucial for G_q- and G_o-Protein Affinity to M₃–R.

To activate mAChRs, we used three different agonists exhibiting different affinities and efficacies toward binding and activation: ACh, its synthetic analog CCh, and the partial agonist arecoline (Are) (42). Thus, we also addressed the question of whether the type of the ligand can affect the affinity of the G protein to the receptor. HEK293T cells were transiently transfected and permeabilized as mentioned above. Single cells were first stimulated with ACh and after its withdrawal, stimulated a second time with CCh or Are (SI Appendix, Fig. S11A). This procedure allows a comparison of the dissociation kinetics of G-protein–receptor complexes in the same cell, meaning equal expression levels of interacting proteins. We observed similar dissociation rates of G_q from ACh-, CCh-, and Are-activated M₃–R (SI Appendix, Fig. S11 A and B). Relatedly, the dissociation kinetics of G_o from M₃–R were not significantly different for all three agonists (SI Appendix, Fig. S11 C and D). Thus, we could conclude that, at least in the case of agonists with moderate to low affinity, the stability of the G-protein–receptor complex is an intrinsic property of the mAChR–G-protein pair and not influenced by the agonist affinity.

Discussion

Understanding the molecular mechanism of selectivity and efficiency of receptor-mediated G-protein activation is one of the key research topics in the field of GPCR physiology and pharmacology. Based on the fact that agonist–receptor–G-protein complexes exhibit the highest stability in absence of nucleotides (43, 44), we quantified the affinity of different G proteins to M₁–, M₂–, and M₃–Rs (22, 28, 45) by means of FRET imaging on permeabilized membranes of single cells. By laminar superfusion of these membranes, we ensured excellent control of agonist and nucleotides (SI Appendix, Fig. S3) and resolved the dynamics of interactions between nucleotide-free G proteins with several different receptors. The lifetime of receptor–G-protein complexes was determined by measuring complex dissociation in response to withdrawal of agonist and was longest for those complexes that contained G-protein subtypes that are known to be activated best by the respective receptor (Figs. 2 and 4 and SI Appendix, Figs. S1 and S10). Our detailed analysis (Fig. 2B vs. Fig. 2 C and D and SI Appendix, Fig. S7) of M₃–R revealed that the lifetime of the nucleotide-free GPCR–G-protein complex (rather than the association kinetics) is the major determinant for the differences in affinity of these complexes (Fig. 3A and SI Appendix, Fig. S6).

Unlike the dissociation kinetics of the nucleotide-free receptor–G-protein complex, the steady-state and on-rate measurements might not exclusively be restricted to complexes after GDP release but also include initially GDP-bound G proteins, although these do not contribute a lot to the overall signal (SI Appendix, Fig. S3 B and D). GDP release is much faster for G_o (46) than for G_q (47), which needs to be considered. We attempted to calculate the k_on of GPCR–G-protein interactions in native cell membranes. The correlation of apparent M₃–R–G_q association kinetics and agonist concentration is consistent with a hyperbolic function (Fig. 2C), allowing us to determine the k_on directly. The resulting value of k_on = 22.4 s⁻¹·mM⁻¹ was remarkably close to the k_on = 14.1 s⁻¹·mM⁻¹ calculated by division of k_off (Fig. 2 A and B) by the EC₅₀ value (Fig. 3A and SI Appendix, Table S1), clearly indicating the applicability of our method. Based on the much faster equilibration kinetics of agonist binding to non-G-protein–bound muscarinic receptors (48), the kinetics of agonist-induced receptor–G-protein complex formation under nucleotide-free conditions is probably a close approximation to the optimal situation of G proteins binding to equilibrated agonist–receptor complexes. In the case of M₃–R–G_o interaction, the correlation of apparent association kinetics of G_o proteins and M₃–R with the agonist concentration is best explained by a two-component hyperbolic function (SI Appendix, Fig. S7). A comparison of the correspondingly determined on-rates with k_on (calculated using Eq. 3) revealed, even with the slower on-rate, a fivefold deviation from the calculated k_on. Possible reasons for this could be effects due to the above-mentioned multistep binding reaction. This might affect the apparent association kinetics of receptors and G proteins and complicates interpretation of the measured k_on. Therefore, at the current research stage, we propose that measuring the off-rates of the complex together with steady-state association gives a more robust quantitative correlate of the affinity of the subtype-specific receptor–G-protein complex.

Our results show that both the steady-state curves of G protein binding to the M₃–Rs as well as the dissociation kinetics of this complex quantitatively reflect M₃–R–G-protein selectivity (Fig. 3). Specifically, the dissociation kinetics of G_o protein from M₃– and M₁–Rs were 13-fold faster in comparison with G_q (Figs. 2 A and B and 4 B and D and SI Appendix, Fig. S10 A, B, and E). Similarly, we observed an approximate 16.5-fold right shift of the concentration–response curves of G_o proteins binding to M₃–R in comparison with G_q (Fig. 3A). A quantitatively similar 6.6-fold difference in coupling efficiency of G_o and G_q to M₃–R, measured in an FRET-based G-protein activation assay in intact cells (Fig. 3C), suggests that the efficacy of coupling to a certain G-protein subtype is indeed reflected in the relative M₃–R–G-protein affinity and can thus be detected by measurement of the lifetime of the G-protein–receptor complex. For moderate- to low-affinity agonists with different efficacy, we could so far not detect differences in the stability of M₃–R–G_q or G_o complexes. An important finding was that the G_q-coupled and evolutionary close M₁– and M₃–Rs exhibited very close ternary complex stabilities, with a very similar pattern for the different G-protein subtypes (Fig. 4 A, B, and D and SI Appendix, Fig. S10 A, B, and E). Remarkably, the stability of the complex of different GPCRs under investigation with their best coupling G protein varied over at least two orders of magnitude, being lowest for M₂–R–G_o protein complexes (lifetime only a few seconds) (SI Appendix, Table S2) and highest for TP-R–G₁₃ and β-AR–Gs complexes (lifetime >200 s) (SI Appendix, Fig. S1 A, C, and E), which likely reflect the parallel evolution of the coupling mechanism (49). In light of the recent advances in understanding the allosteric effect of G protein binding (or mimicking nanobodies) to the M₂–R–agonist interaction, specifically the closure of a tyrosine lid on the extracellular side of the agonist exit path (13, 50, 51), our results that M₂–R–G_o interaction is very short lived may be somewhat unexpected. However, they are not contradictory to published results, as the lifetime of M₂–R–G_o complexes was not addressed in previous studies. The reliability of the method that we developed was also confirmed by testing other G-protein classes. For instance, we were unable to detect G_s or G₁₃ binding to M₃–R or M₁–R under nucleotide-depleted conditions (Fig. 2A and SI Appendix, Fig. S10B) or to observe activation of these G proteins measured by means of FRET (SI Appendix, Fig. S2).

Furthermore, we screened other G_i family proteins, and although we could not observe significant differences in the affinities of G_i proteins to M₃– or M₁–Rs, M₂–R did show a higher specificity for G_o and G_i2 over G_i1 and G_i3 (Fig. 4C and SI Appendix, Fig. S10 D and F). This distinct affinity profile of GPCR is also supported by recently delineated GPCR “fingerprints” determined as the efficiency of G-protein activation by different GPCRs (52). Moreover, recent computational and evolutionary studies assume that the selectivity mechanism of GPCRs is likely disclosed on the G-protein level (15, 49). Thus, we conclude that receptor–G-protein affinity represents a major determinant for receptor–G-protein subtype selectivity. We could essentially attribute the observed differences in the affinity of G_o versus G_q toward active M₃–Rs to differences in the lifetime of the complex, suggesting that the apparent on-rate of complex formation is less important for defining coupling selectivity.

The potential and distinctive advantage of our FRET-based nucleotide-free method over biochemical assays is the ability to perform experiments in a regular membrane environment. By excluding the steps of protein purification and protein reconstitution (53, 54), a remarkably high degree of signal specificity is still exhibited. This is particularly important, as G proteins are well-known to be very sensitive to detergents (55). Furthermore, this method is easily accessible to all GPCRs and G proteins and can be used as a reliable way to quantify GPCR–G-protein specificity.

Materials and Methods

Chemicals.

DMEM, FCS, PBS, penicillin/streptomycin, l-glutamine, and trypsin-EDTA were from Biochrom. Saponin was purchased from AppliChem. All other substances were purchased from Sigma-Aldrich.

Plasmids.

cDNAs for Gα_q; Gα_q-YFP (56); Gβ₁-WT; Gγ₂-WT; Gα_o-YFP (C351I) (57); Gα_i1 (C351I)-, Gα_i2-, Gα_i3- (58), and Gα_o-WT (59); CFP-Gγ₂ (37); and M₃–R–YFP (48) were described previously. The M₃–R was obtained from the Missouri S&T cDNA Resource Center. M₃–R–mTurquoise fluorescent protein was cloned analogously to M₃–R–YFP.

Cell Culture and Transfection.

All experiments were carried out in HEK293T cells (a gift from Dr. Martin Lohse, Institute of Pharmacology, Würzburg, Germany). Cells were cultured in high-glucose DMEM supplemented with 10% FBS, 2 mM l-glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin at 37 °C and 5% CO₂. Cells were transiently transfected with 0.5 µg DNA (6-cm dish) of each particular mAChR-YFP, 1.5 µg DNA Gα-WT of interest (if not otherwise indicated), 0.5 Gβ₁-WT, 0.2 CFP-Gγ₂, and 0.3 pcDNA3 (if not otherwise indicated) using Effectene Transfection Reagent according to the manufacturer’s instructions (Qiagen). For the G-protein activation, HEK293T cells were transfected with (micrograms DNA, 6-cm dish) 0.5 µg M₃–R–WT, 0.8 µg Gα-YFP of interest, 0.5 µg Gβ₁-WT, and 0.2 µg CFP-Gγ₂. Experiments were performed 48 h after transfection at room temperature on cells, which were plated into six-well plates with 25-mm coverslips (cover glasses were preincubated 30 min with poly-l-lysine) 1 d after transfection.

Permeabilization Procedure.

Coverslips with transiently transfected HEK293T cells were fixed in a microscope chamber and washed once with external buffer (137 mM NaCl, 5.4 mM KCl, 10 mM Hepes, 2 mM CaCl₂, 1 mM MgCl₂, pH 7.3). To permeabilize cells, coverslips were incubated for 2 min with 0.05% saponin and afterward washed five times with internal buffer (100 mM K⁺-aspartate, 30 mM KCl, 10 mM Hepes, 5 mM EGTA, 1 mM MgCl₂, 10 mM NaCl, pH 7.35.), inducing the depletion of GTP and GDP (60).

Single-Cell FRET Imaging.

The FRET measurements were performed on single cells selected for membrane staining of YFP and CFP fluorescence. The round shape of cells was taken as an indicator of proper permeabilization. Dual-emission imaging of YFP and CFP of a single cell (or its membrane) was performed as previously described in the work by Milde et al. (36) and modified as described in SI Appendix, SI Methods. Individual traces are shown as either absolute or relative YFP/CFP emission ratio changes. Absolute changes in YFP/CFP emission ratio were calculated as the differences in average values of the last 10 s before or after an event. The relative change of F_YFP/F_CFP was calculated by normalization to the maximum peak after application of saturating agonist concentration relative to the G-protein– state determined by application of GTPγS. As a measure of the affinity of the G-protein–receptor complex, dissociation kinetics of the complex were measured under nucleotide-free conditions in response to agonist withdrawal. Resulting data of the offset kinetics (decrease in FRET) were fitted by a monoexponential function.

Multiple-Cell FRET Imaging.

The efficiency of G-protein activation by the receptor of interest was determined by means of a multiple-cell FRET assay in intact cells transiently transfected as previously described in the work by Frank et al. (59). Transfected HEK293T cells cultivated in 12-well strips (TC; 96-well format; Greiner Bio-One) were washed with and maintained in external buffer A (137 mM NaCl, 5.4 mM KCl, 1.0 mM MgCl₂, 2.0 mM CaCl₂, 10 mM Hepes, 2% Brilliant Black) at 32.5 °C. The final volume was 275 µL per well. Injections of test compounds (agonists and antagonists) were made with a volume of 22 µL added gradually with gentle up and down movements of the 12-channel injecting unit equipped with 50-mm needles (1.2-mm diameter), resulting in a 13.5-fold dilution of stock solutions. Alternatively, cells were stimulated with an overflow apparatus, allowing almost immediate (<3 s) change of the standard buffer to induce receptor stimulation (Fig. 3D). Details for the optical equipment are given in SI Appendix, SI Methods.

Data Processing.

Fluorescence intensities were acquired using the imaging software NIS-Elements advanced research (Nikon Corporation). Values are given as means ± SEM of n experiments. Other programs used for the analysis are given in SI Appendix, SI Methods.