To measure the rearrangement or dissociation of the G_i/o proteins upon receptor activation in live cells and neurons, we used a series of BRET-based biosensors. They are based on the use of a luciferase as an energy donor inserted in the Gα subunit and the fluorescent protein Venus as an energy acceptor attached to the Gγ subunit, as reported for previous BRET- and FRET-based G protein sensors^23–25. To monitor these sensors in conditions close to the physiological ones, in our approach, the cells were cotransfected only with the luciferase-tagged Gα subunit and Venus-tagged Gγ subunit (^VenusGγ), while the endogenous Gβ subunits were used (Fig. 1a). In these experiments, the amount of Gβγ complexes that can produce a BRET signal with Gα is expected to be limited by the endogenous level of Gβ subunits (Supplementary Fig. 1a) that are able to form a complex with the ^VenusGγ. We used mostly the mouse cerebellum granule neurons (CGNs) that constitute the largest homogenous neuronal population of the mammalian brain³⁴.Fig. 1

a Scheme of the BRET-based sensors to detect the activity of the endogenous GPCR in neurons. b Scheme of the Gα_i/o^Nluc and ^VenusGγ₂ constructs co-transfected in the neurons. Sequence of the linker (L) before and after Nluc was indicated. Gγ was fused with Venus in the N-terminal. c Composition of G protein sensors made of the different Gα_i/o^Nluc and ^VenusGγ₂. d Kinetics of the BRET signal between the indicated Gα constructs (Gα_i1^Nluc or Gα_i1^Rluc8) and ^VenusGγ₂ in CGNs. Buffer or baclofen (100 μM) were injected at the indicated time (arrow). The data are representative of BRET ratios from five independent experiments. e Representative fluorescent images of Venus and DAPI in CGNs transfected with Gα_oA^Nluc and ^VenusGγ₂ or in HEK293 cells transfected with Gα_oA^Nluc, G_β1 and ^VenusGγ₂ from three independent experiments. Scale bar: 20 μm. f Kinetics of the BRET signal between Gα_i1^Nluc and ^VenusGγ₂ in CGNs. Buffer or baclofen (100 μM) or competitive antagonist CGP64213 (10 μM) were injected at the indicated time. The data are representative of the mean ± SEM of BRET ratios from three independent experiments. The raw data and p-values are available in source data provided as a Source Data file.

Our sensors rely on the use of a small luciferase, the Oplophorus nanoluciferase (Nluc; 19 kDa) that produces a more intense and sustained luminescence signal than other commonly used luciferases from Renilla luciferase (Rluc and Rluc8; 36 kDa)³⁵. Nluc was inserted in the helical domain of different Gα_i/o subunits similarly to previously reported G protein BRET sensors with Rluc^23,24 and Nluc³⁶, and the constructs were named Gα^Nluc (Fig. 1b). The Nluc was inserted after residue 91 (for Gα_i1, Gα_i2, Gα_i3, Gα_oA and Gα_oB), or 113 (for Gα_z) (Fig. 1c). Of note, Gα_oA has 94% sequence identity with Gα_oB and around 70% with Gα_i1, Gα_i2 and Gα_i3³⁷. To facilitate the insertion of Nluc, a short and flexible linker sequence (SGGGGS) was added at the N- and C-terminal ends of Nluc. For BRET signal measurement, Venus was fused to the N-terminus of Gγ₂ subunit with the amino acid sequence KLGT serving as a linker, and the construct is named ^VenusGγ₂ (Fig. 1b). To show the advantage of using Nluc over Rluc8 in measuring agonist-induced BRET change in neurons, Rluc8 was inserted at the same position as Nluc in Gα_i1 (named Gα_i1^Rluc8) and Gα_oA (named Gα_oA^Rluc8). These Gα_i1^Rluc8 and Gα_oA^Rluc8 constructs are highly similar to the constructs reported for the Gα protein of TRUPATH BRET sensors (Supplementary Fig. 1b). The Gα_oA^Nluc construct showed much higher luminescence intensity at emission 480 nm compared with the subunits Gα^Rluc8 in CGNs, for the same amount of cDNA co-transfected with ^VenusGγ₂ (Supplementary Fig. 1c). In addition, the basal BRET ratio of the constructs Gα_i1^Nluc and Gα_oA^Nluc co-transfected with ^VenusGγ₂ was also more stable over time, compared to the sensors that used subunits Gα^Rluc8 (Fig. 1d and Supplementary Fig. 1d). We also verified that the co-expression of the constructs Gα^Nluc + ^VenusGγ₂ for the different Gα subunits did not change the expression level of endogenous Gβ subunits (Supplementary Fig. 1e). Altogether, the G protein Nluc-biosensors are expected to produce a much improved signal-to-noise ratio for BRET measurement in neurons.

We then validated our different G_i/o protein sensors with the GABA_B receptor as a prototype G_i/o-coupled receptor, which is abundantly expressed in many types of neurons³⁸, including in CGNs³⁹. Interestingly, despite a low percentage of transfected neurons compared with HEK293 cells (Fig. 1e), the GABA_B receptor specific agonist baclofen largely decreased the BRET signal with the sensor Gα_i1^Nluc + ^VenusGγ₂ (Fig. 1d, Supplementary Fig. 2a) and Gα_oA^Nluc + ^VenusGγ₂ (Supplementary Fig. 1d, 2b), while the change was barely detectable with Gα_i1^Rluc8 + ^VenusGγ₂ (Fig. 1d) and Gα_oA^Rluc8 + ^VenusGγ₂ (Supplementary Fig. 1d). In kinetics experiments, baclofen induced a rapid and strong decrease of BRET signal in CGNs, that was reversed by the GABA_B receptor competitive antagonist CGP64213 for both the G_i1 (Fig. 1f) and G_oA (Supplementary Fig. 2c) sensors. As a control, no change of BRET signal was measured when Gα_i1^Nluc was expressed in the absence of ^VenusGγ₂ or with Venus (Supplementary Fig. 2d). Similarly, no change of BRET signal was measured with the dominant negative mutants Gα_i1^Nluc-S47A⁴⁰ or Gα_i1^Nluc-G202T⁴¹ (Supplementary Fig. 2e) when they are co-expressed with ^VenusGγ₂. This is consistent with the unability of these mutants to exchange their GDP for GTP. We also verified that the co-expression of Gβ with the Gα and Gγ₂ was not required for a large BRET change signal induced by baclofen (Supplementary Fig. 2f). Altogether, these results validated the use of the Gα_i1^Nluc and Gα_oA^Nluc -based BRET sensors in neurons.

These G_i/o protein Nluc sensors were highly sensitive even though only about 2% of CGNs expressed the sensor as measured by the fluorescence of the transfected ^VenusG_γ2. This low level of transfection is consistent with previous data reported with CGNs⁴², even though any transfected cell is expected to overexpress the sensors thanks to the strong promotor. The transfected neurons displayed normal morphology with dendrites and branches (Fig. 1e). It indicates that the Gα_i/o protein Nluc-sensors were sensitive enough to monitor endogenous GPCR activation in a small number of neurons. In addition, they are compatible with measurements in 96-well plates. The amount of cDNA to be transfected in CGNs was optimized for each sensor (Supplementary Fig. 2g). Finally, our sensors are also compatible with the measurement of the activation of endogenous GPCRs expressed in the HEK293 cells such as the lysophosphatidic acid receptor (Supplementary Fig. 3a). In these experiments, the G_i2 showed a weak response than those generated by the other G_i/o sensors (Supplementary Fig. 3a), and this is unrelated to differences in expression levels of these sensors (Supplementary Fig. 3b).

We then compared the agonist-induced BRET signals obtained with the same Gα_i/o sensor in the two different cell types, CGNs and HEK293 cells (Fig. 2a). Conditions were optimized such that the GABA_B receptor expression level in HEK293 cells was similar to that in CGNs (Fig. 2b). Since the basal BRET signal in the absence of agonist can be different between the G protein sensors (Supplementary Fig. 3c), the signal was expressed as the agonist-induced change in BRET ratio expressed as percentage of the basal signal. The results showed that in CGNs, the BRET changes induced by baclofen are significantly different from those measured in HEK293 cells except for the two G_o sensors (Fig. 2a). Indeed, a larger BRET signal is measured in CGNs with G_i1 and G_i3, and a large signal is observed with Gz in HEK293 cells while absent in CGNs. The differences observed are related neither to the expression levels of the Gα^Nluc sensor components (Fig. 2c) nor to the endogenous Gα_i/o mRNA expression levels (Fig. 2d). Though lower signal was observed in G_i2 in CGNs, the difference of baclofen response can be detected in dose-dependent manner (Supplementary Fig. 3d). Similar pEC₅₀ values were measured in three G_i subtype sensors and two G_o subtype sensors in CGNs upon baclofen activation (Supplementary Fig. 3d). Altogether, the data showed different abilities of the GABA_B receptor to generate responses by the different G_i/o subtypes in transfected HEK293 cells and in CGNs, then revealing the importance of studying GPCRs in their native environment.Fig. 2

a Baclofen-induced change in BRET ratio between Gα^Nluc and ^VenusGγ₂ in CGNs or transfected HEK293 cells for the indicated G_i/o proteins, and expressed as percentage of the basal signal ((BRET_basal-BRET_agonist/BRET_basal) x 100). CGNs were co-transfected with ^VenusGγ₂ (50 ng) and the Nluc-tagged Gα_i1 (25 ng), Gα_i2 (75 ng), Gα_i3 (75 ng), Gα_oA (25 ng), Gα_oB (25 ng) or Gα_z (25 ng), while HEK293 cells were co-transfected with GB1 (9 ng), GB2 (12 ng), Gβ₁ (10 ng)_, ^VenusGγ₂ (10 ng) and the Nluc-tagged Gα_i1 (1 ng), Gα_i2 (3 ng), Gα_i3 (3 ng), Gα_oA (1 ng), Gα_oB (1 ng) or Gα_z (1.5 ng) per well in 96-well plate. b Expression of GB1 subunit of GABA_B receptor in the cell membrane of CGNs, mock HEK293 cells, and HEK293 cells transfected with indicated amount of GB1 and GB2, detected by western blotting. Values are mean ± SEM normalized as fold of GB1 expression in CGNs from four independent experiments. c Expression of the Nluc-tagged Gα transfected in CGNs or in HEK293 cells in (a), measured by the luminescent signal at 480 nm. The values of individual well are shown. Values are mean ± SEM from four biologically independent experiments each performed in triplicates or quadruplicate in (a) and (c). Data in (a) are analysed using unpaired t-test (two-tailed) to determine significance. Data in (c) are analysed using one-way analysis of variance (ANOVA) with a Dunnett’s post-hoc multiple comparison test to determine significance (compared with G_i1 group). ****p < 0.0001, ***p < 0.001, **p < 0.01 and *p < 0.05 and not significant (ns). d Expression of the genes encoding indicated Gα in CGNs (mouse species: Gnai1, Gnai2, Gnai3, Gnao-1, Gnao-2, Gnaz) and HEK293 cells (human species: GNAI1, GNAI2, GNAI3, GNAO-1, GNAO-2, GNAZ). Values were determined by qRT-PCR and mean ± SEM normalized to Rplp0 (mouse) or RPLP0 (humans) from three biologically independent experiments each performed in quadruplicate. The values of individual sample are shown. The raw data and p-values are available in source data provided as a Source Data file.

We have further validated these G protein sensors by measuring the potency of different orthosteric GABA_B receptor ligands (Fig. 3a) in both CGNs and HEK293 cells. Similar potencies were obtained for the G_i1 and G_oA for baclofen and other agonists, and for the antagonist CGP64213 in CGNs (Fig. 3b–f, Supplementary Table 1). The GABA_B receptor agonists APPA and SKF 97541 were more potent than GABA and baclofen in both CGNs (Fig. 3c–f) and HEK293 cells (Supplementary Fig. 4a), which was consistent with previous findings using the GTPγS assay in brain tissues²¹. pEC₅₀s obtained in CGNs were correlated with those measured in HEK293 cells (Fig. 3e, f), even though they were more than 10 times higher in HEK293 cells, except for baclofen where the difference is lower (Supplementary Table 1, Fig. 3e, f). Similar results were obtained in the hippocampal neurons and cortical neurons (Supplementary Fig. 4b–d and Supplementary Table 1). These differences in baclofen potencies observed between transfected HEK293 cells and native neurons were not due to the difference in receptor expression. Indeed, when the expression level GABA_B receptors was similar to that in CGNs (Fig. 2b), the baclofen pEC₅₀ was similar to that observed with higher expression levels, and still significantly different from that measured in CGNs (Fig. 3g).Fig. 3

a Scheme of the GABA_B receptor, made of GB1 and GB2 subunits, in CGNs and its various specific ligands (agonist, red; antagonists, blue; PAM, black). b Change of BRET signal induced by the GABA_B receptor antagonist CGP64213 in presence of 20 μM baclofen (EC₈₀) or different doses of baclofen in CGNs measured by G_i1 or G_oA sensor. c, d Change of BRET ratio induced by various GABA_B receptor agonists measured by G_i1 (c) or G_oA (d) sensor. e, f Correlation of the agonist potencies (pEC₅₀) between HEK293 cells and CGNs determined by the G_i1 (e) or G_oA (f) BRET sensors. Dotted line is the correlation of pEC₅₀ determined with HEK293 cells. Red line is fit of pEC₅₀ in CGNs with the same slope as the dotted line. g The pEC₅₀ of baclofen in CGNs and transfected HEK293 cells with indicated amount of GB1 and GB2 measured by G_i1 or G_oA sensor. Data are mean ± SEM from at least three biologically independent experiments each performed in triplicate and analysed using one-way ANOVA with a Dunnett’s post-hoc multiple comparison test to determine significance. The n number of G_i1 and G_oA group for CGNs, and HEK293 cells transfected with the indicated amount of GB1 and GB2 are 10, 4, 3, and 4, 3, 3, respectively. ****p < 0.0001, **p < 0.01 and not significant (ns). h Change of BRET ratio induced by baclofen with or without indicated concentrations of the GABA_B PAM Rac BHFF in transfected HEK293 cells and CGNs measured by G_i1 sensor. Data are normalized to maximum of baclofen response and mean ± SEM from at least three biologically independent experiments each performed in triplicate in (b–d) and (h). b n = 3; c n = 6; d n = 4; h CGNs n = 3, HEK293 cells n = 4. The raw data and p-values are available in source data provided as a Source Data file.

We have also validated our sensors for the analysis of positive allosteric modulators (PAMs). PAMs can synergistically enhance the activity of native neurotransmitter GPCRs in neurons⁴³ and have a greater potential for drug development with less side effect⁴⁴, including for the GABA_B receptor⁴⁵. It was recently demonstrated that these PAMs bind at the interface of the transmembrane domains of this dimeric receptor^46–48 (Fig. 3a). Here we showed that Rac BHFF increased strongly the potency of baclofen-induced G_i1 (Fig. 3h) and G_oA protein (Supplementary Fig. 5a) response in CGNs. However, in contrast to what is observed in HEK293 cells^48,49 (Fig. 3h and Supplementary Fig. 5a–b), Rac BHFF alone (10 μM) did not increase the net BRET signal in CGNs (Fig. 3h and Supplementary Fig. 5a). It is nicely illustrated by concentration responses curves of the Rac BHFF obtained with transfected HEK293 cells but not so efficiently with neurons (Supplementary Fig. 5c). It suggests that the Rac BHFF agonist activity observed with the recombinant GABA_B receptor is much higher than that measured for the receptor in its native environment. This slight agonist effect of Rac BHFF in CGNs is probably not due to the lower GABA_B receptor expression in neurons compared to HEK293 cells. Indeed, the agonist activity of Rac BHFF can still be observed in HEK293 cells that expressed GABA_B receptor at a similar level to that found in CGNs (Figs. 2b, 3h, Supplementary Fig. 5a).

Finally, we took advantage of our sensors to measure the constitutive activity⁵⁰, in the absence of agonist, of the GABA_B receptor that is usually observed in recombinant systems^{47,49,51–55}. This constitutive activity was reversed by the competitive antagonists known for its inverse agonist activity^49,51,55,56. Accordingly, the competitive antagonist CGP54626 increased the basal BRET signal between Gα and Gβγ for both the G_i1 and G_oA sensors in the HEK293 cells (Supplementary Fig. 5d), as previously reported⁴⁹. It is consistent with the properties of this inverse agonist to favour the Gαβγ heterotrimer by stabilizing the inactive conformation of the receptor. Interestingly, this increase of BRET signal was not observed in the CGNs (Supplementary Fig. 5d), indicating the native GABA_B receptor has no or a very low constitutive activity that cannot be observed with our sensor. It is consistent with the weak intrinsic agonist activity of Rac BHFF in neurons.

Altogether, our results show that GABA_B receptor ligands have different potencies and efficacies in the cell lines compared to neurons, which cannot be explained by a difference in GABA_B receptor expression levels between the two cell types.

We then examined the activities of other endogenous G_i/o-coupled GPCRs in the CGNs. First, the agonist CP 55940 that actives both the cannabinoid receptor type 1 (CB1) and 2 (CB2) induced a strong change in both G_i1 and G_oA BRET signal as observed for the GABA_B receptor (Fig. 4a). It was expected that it is due to the activation of CB1 since this receptor is highly expressed in cerebellum³³ compared to CB2 that is mainly expressed at the periphery⁵⁷. When testing other GPCR agonists, a strong or significant change of BRET signal was measured with different ligands used at saturation concentrations for their expected receptors: brimonidine (also named UK-14304), pramipexole, eletriptan and carbachol known to activate α₂ adrenoceptor⁵⁸, D_2/4 receptor⁵⁹, 5-hydroxytryptamine receptor 1B/1D⁶⁰ and the M_2/4 muscarinic receptors⁶¹, respectively. In contrast, agonists of the platelet-activating factor (PAF) receptor and δ/μ opioid receptor DAMGO failed to induce a BRET signal change (Fig. 4a). Dose-response of brimonidine, CP 55940, pramipexole and carbachol further confirmed the specific activation of these endogenous G_i/o-coupled receptors (Fig. 4b). Altogether, our data correspond to the receptors for which higher amount of mRNA were detected in these cultured CGNs at the developmental stage investigated⁶².Fig. 4

a Net BRET signal of the G_i1 and G_oA sensors in CGNs induced by specific agonists of various G_i/o-coupled GPCRs using a saturating concentration of the indicated ligands (PAF, 10 μM; DAMGO, 100 μM; eletriptan, 100 μM; brimonidine, 10 μM; carbachol, 100 μM; pramipexole, 100 μM; CP 55940, 20 μM and baclofen, 100 μM). Values are mean ± SEM from at least biologically independent experiments each performed in triplicate. The n number of the indicated treatments in G_i1 and G_oA are 6, 4, 5, 4, 6, 4, 6, 7, 8, and 8, 6, 4, 6, 6, 6, 6, 6, 7, 8, respectively. Data are analysed using one-way ANOVA with a Dunnett’s post-hoc multiple comparison test to determine significance (compared with no treated group). ****p < 0.0001, *p < 0.05 and not significant (ns). b Dose-response of the agonists brimonidine, CP 55940, carbachol and pramipexole for the α₂ adrenoceptor, CB1, muscarinic M_2/4 and dopamine D_2/4 receptors, respectively, in CGNs co-transfected with Gα_i1^Nluc and ^VenusGγ₂ or Gα_oA^Nluc and ^VenusGγ₂. Data are normalized to maximum response of each compound and mean ± SEM from at least three biologically independent experiments each performed in triplicate. G_i1, brimonidine (n = 3), CP 55940 (n = 4), carbachol (n = 4) and pramipexole (n = 4); G_oA, brimonidine (n = 3), CP 55940 (n = 3), carbachol (n = 4) and pramipexole (n = 3). The raw data and p-values are available in source data provided as a Source Data file.

Next, using α₂ adrenoceptor we show that our G protein sensors can determine ligand efficacies, by evaluating the full agonist brimonidine and several reported partial agonists (clonidine, oxymetazoline, tizanidine and xylazine)⁶³. In contrast, brimonidine exhibited the highest response of G_i1 and G_oA protein in CGNs, while others showed less response (Fig. 5a). However, the discrimination between brimonidine and other agonists was not significantly observed in HEK293 cells transfected with α_2A adrenoceptor (α_2AAR) (Fig. 5b). It was not due to the high level of expression of α_2AAR, since no discrimination between brimonidine and other agonists was observed in HEK293 cells that expressed α_2AAR at a similar level to that found in CGNs (Fig. 5c). In these conditions (15 ng of α_2AAR cDNA/well), the brimonidine pEC₅₀ was similar to that observed with higher expression levels (Supplementary Fig. 6a), and still significantly different from that measured in CGNs (Fig. 5d), as observed previously for the GABA_B receptor (Fig. 3g). Finally, the data showed a significant difference in the G protein responses mediated by the different G_i/o subtypes triggered by the recombinant α_2AAR in HEK293 cells and the endogenous one in CGNs (Fig. 5e and Supplementary Fig. 6b). Of note, similarly to experiments above with GABA_B receptor (Fig. 2c), the expression of the different Gα^Nluc proteins was within a factor of three in each type of cell (Supplementary Fig. 6c).Fig. 5

a Effect of the indicated α₂AR agonists on the net BRET of the G_i1 and G_oA sensors in CGNs co-transfected with Gα_i1^Nluc and ^VenusGγ₂ or Gα_oA^Nluc and ^VenusGγ₂ (amounts of cDNA as in Fig. 2a). b Effect of the indicated α₂AR agonists on the net BRET of the G_i1 and G_oA sensors in HEK293 cells co-transfected with the mouse α_2AAR (10 ng, 20 ng or 50 ng), and Gβ_1, ^VenusGγ₂ and Gα_i1^Nluc or Gα_oA^Nluc as in Fig. 2a. Saturating concentrations of brimonidine (10 μM), oxymetazoline (100 μM), xylazine (50 μM), clonidine (100 μM), tizanidine (100 μM) were used in (a, b). c Detection of α₂AR in cell membranes of CGNs, mock-transfected HEK293 cells and HEK293 cells transfected with indicated amount of α₂AR, by western blotting. Values are mean ± SEM normalized as fold of α₂AR expression in CGNs from four independent experiments. d The pEC₅₀ of brimonidine in CGNs (n = 3) and transfected HEK293 cells (n = 4) with indicated amount of α₂AR measured by G_i1 or G_oA sensor. e Percentage of change in BRET ratio between Gα^Nluc and ^VenusGγ₂ induced by brimonidine between Gα^Nluc and ^VenusGγ₂ in HEK293 cells (n = 3) (α₂AR: 15 ng/well per 96-well plate) or CGNs (n = 4) for the indicated Gα_i1, Gα_i2, Gα_i3, Gα_oA, Gα_oB or Gα_z sensors (amounts of cDNA as in Fig. 2a). Values are mean ± SEM from at least three biologically independent experiments each performed in triplicate or quadruplicate in (a, b, d, e). a, n = 4; b, n = 3; (d, e), CGNs, n = 3; HEK293 cells, n = 4. Data are analysed using one-way ANOVA with a Dunnett’s post-hoc multiple comparison test to determine significance (compared with brimonidine in a, b). Data are analysed using unpaired t-test (two-tailed) in (e). ****p < 0.0001, ***p < 0.001, **p < 0.01 and not significant (ns). The raw data and p-values are available in source data provided as a Source Data file.

Our results demonstrated that G_i/o protein Nluc-biosensors can be widely applied to different endogenous G_i/o-coupled GPCRs and evaluate the ligand efficacies and G protein coupling profile in neurons. They also further revealed the importance of studying a GPCR in its native environment.

Finally, we took advantage of the major change in BRET signal observed for the cannabinoid receptor to investigate the G protein response in CGNs. While level of expression of the CB1 receptor in the two systems was similar in western blotting experiments (Fig. 6a), a higher CB1 receptor-mediated G protein response was observed with G_i1 and G_i3 only, with the native receptor in CGNs compared to the recombinant one in HEK293 cells (Fig. 6b, Supplementary Fig. 7a). Similar responses were observed with the other G_i/o subtype sensors between two types of cells. Again, as with the GABA_B receptor (Fig. 2b) and α_2AAR (Supplementary Fig. 6c), the expression of the different Gα^Nluc proteins was within a factor of three in each type of cell (Supplementary Fig. 7b). And a small overexpression of the CB1 receptor by transfection of CGNs with the recombinant CB1 receptor gave similar results (Supplementary Fig. 7c, d). Interestingly, our G protein sensors revealed a higher potency of CP 55940, Win 55,212-2 and Bay 59-3074 in CGNs compared with transfected HEK293 cells for both G_i1 and G_oA response (Fig. 6c, d and Supplementary Table 2). It is different from GABA_B and α₂AR, which showed higher agonist potencies in transfected HEK293 cells. Of note, when taking Win 55,212-2 as a reference ligand, Bay 59-3074 showed G_oA bias in both CGNs and HEK293 cells with a bias factor of 1.6 and 1.3 respectively. CP 55940 showed a G_i1 bias in CGNs and a G_oA bias in HEK293 cells with a bias factor of 1.6 and 1.4 respectively, indicating a difference between G_i1 and G_oA proteins revealed by CP 55940 (Fig. 6c, d and Supplementary Tables 2 and 3).Fig. 6

a Detection of CB1 receptor in cell membranes of CGNs, mock-transfected HEK293 cells and HEK293 cells transfected with mouse CB1 cDNA (80 ng/well in 96-well plate), by western blotting. Values are mean ± SEM normalized as fold of CB1 expression in CGNs from three independent experiments. b Percentage of change in BRET ratio between Gα^Nluc and ^VenusGγ₂ induced by CP 55940 in HEK293 cells (CB1 cDNA; 80 ng/well per 96-well plate) or CGNs for the indicated Gα_i1, Gα_i2, Gα_i3, Gα_oA, Gα_oB or Gα_z sensors (amounts of cDNA as in Fig. 2a). Values are mean ± SEM from four biologically independent experiments each performed in triplicate. Data are analysed using unpaired t-test (two-tailed) to determine significance. ****p < 0.0001, *p < 0.05 and not significant (ns). c, d Change of BRET signal between Gα^Nluc and ^VenusGγ₂ induced by the indicated CB1 receptor agonists in HEK293 cells and CGNs measured by G_i1 and G_oA sensors. The transfection was the same as in (b). Inset, correlation of the agonist potencies (pEC₅₀) between HEK293 cells and CGNs determined by the G protein sensors. Dotted lines are the correlation of pEC₅₀ determined with HEK293 cells. Red lines are the fit of pEC₅₀ in CGNs with the same slope as the dotted lines. Data are normalized to maximum CP 55940 response and are mean ± SEM from at least three biologically independent experiments each performed in triplicate (HEK293 cells: G_i1, CP 55940 (n = 5), Win 55,212-2 (n = 5) and Bay 59-3074 (n = 4); G_oA, CP 55940 (n = 5), Win 55,212-2 (n = 4) and Bay 59-3074 (n = 4). CGNs: G_i1, CP 55940 (n = 4), Win 55,212-2 (n = 4) and Bay 59-3074 (n = 4); G_oA, CP 55940 (n = 4), Win 55,212-2 (n = 4) and Bay 59-3074 (n = 3)). The raw data and p-values are available in source data provided as a Source Data file.

The influence of Gβγ subunit composition on G protein coupling^64,65 was largely reported even though the underlying molecular mechanism is largely unknown. Therefore, we have compared sensors composed by Gα^Nluc and different Gγ including ^VenusGγ₂, ^VenusGγ₈ or ^VenusGγ₉, which are well-used in the previously reported G protein BRET sensors^25,36. ^VenusGγ₈ or ^VenusGγ₉ belong to two different groups of Gγ subunits (Supplementary Fig. 8). Gγ₈, like Gγ₂, is well expressed in some regions of the brain, in contrast to Gγ₉ that seems restricted to retina⁶⁶. Interestingly, a major difference in G protein response profile of the GABA_B receptor and α₂ adrenoceptor was measured depending on the Gγ subunits used (Fig. 7a–c and Supplementary Fig. 9a–d), that clearly shows the impact of the Gγ subunit in these GPCR-mediated responses.Fig. 7

a, b Percentage of change in BRET ratio between Gα^Nluc and ^VenusGγ₈ (a) or ^VenusGγ₉ (b) induced by baclofen, brimonidine and CP 55940 in CGNs for the indicated G_i/o proteins. CGNs were co-transfected with the indicated ^VenusGγ and Nluc-tagged Gα_i1, Gα_i2, Gα_i3, Gα_oA, Gα_oB or Gα_z. Values are mean ± SEM from biologically independent experiments (baclofen and brimonidine, n = 5; CP 55940, n = 3) each performed in triplicate. c Scheme illustrating the difference in G_i/o (G_i1, G_i2, G_i3, G_oA, G_oB and Gα_z) and indicated Gγ protein (Gγ₂, Gγ₈ and Gγ₉) responses in CGNs for the receptors GABA_B, CB1 and α₂AR in presence of the indicated agonists. For each receptor and G_i/o sensor, the size of the circle is the percentage of change in BRET ratio upon the agonist stimulation, for GABA_B (Figs. 2a and a, b), α₂AR (Figs. 5e and a, b) and CB1 (Figs. 6b and a, b) normalized to the G_i/o sensor that has the highest response: for Gγ₂, G_i1 for both the GABA_B and CB1, and Gα_z for α₂AR; for Gγ₈, G_oA for GABA_B and α₂AR and G_i1 for CB1; and for Gγ₉, G_i1 for GABA_B and CB1, and Gα_z for α₂AR. For Gγ8, the empty circle indicates an increase in the BRET ratio in contrast to other conditions where a decrease BRET ratio is measured. The raw data and p-values are available in source data provided as a Source Data file.

GABA (Cat. A2129), CP 55940 (Cat. C1112) and carbachol (PHR1511) were purchased from Sigma-Aldrich (Shanghai, China). R-baclofen (Cat. 0796), SKF 97581 (Cat. 0379) and CGP54626 (Cat. 1088) were obtained from Tocris Biosciences (Shanghai, China). 3-APPA (Cat. ab120329) was purchased from Abcam (Shanghai, China). Win 55,212-2 (T4458), Bay 59-3074 (T3699), DAMGO (T4351), eletriptan HBr (T0216), pramipexole 2HCl monohydrate (T6951), brimonidine tartrate (T6422), tizanidine hydrochloride (T0290), oxymetazoline hydrochloride (T0252), xylazine (T7046), clonidine hydrochloride (T1247), and 1-oleoyl lysophosphatidic acid sodium (T21654) were purchased from TargetMol (Shanghai, China). PAF C-16 (74389-68-7) was purchased from Santa Cruz Biotechnology (Shanghai, China). CGP64213 was a gift from Prof. Fajun Nan (Shanghai Institute of Materials Medicine, China). Coelenterazine h (Cat. S2001) and furimazine (Cat. N1120) were purchased from Promega (Madison, WI, USA).

The coding sequences of Nluc and Rluc8 were PCR-amplified and inserted into the coding sequence of human Gα_i1, Gα_i2, Gα_i3, Gα_oA and Gα_oB (between residues 91 and 92) and Gα_z (between residues 113 and 114) in pcDNA3.1, using the flexible linkers SGGGGS and SGGGEF, before and after the sequence of Nluc, as shown in Supplementary Figs. 10c, d and 12. The pRK5 plasmid encoding the rat GABA_B subunits (HA-tagged GABA_B1 (GB1) and Flag-tagged GABA_B2 (GB2)) and pcDNA3.1 encoding the human Gβ₁ and human ^VenusGγ₂ were provided by the Institut de Génomique Fonctionnelle (Montpellier, France). The cDNA of mouse α_2AAR, mouse CB1, human Gγ₈ and human Gγ₉ were bought from Miaoling Bio (Wuhan, China) and inserted in pcDNA3.1. Venus was inserted at the N-terminus of Gγ₂, Gγ₈ and Gγ₉ between the HindIII and KpnI restriction enzymatic sites and the linker KLGT between Venus and the Gγ subunits was used. The schematic representation of the constructs (Supplementary Fig. 10) and their protein sequences (Supplementary Figs. 11–13) were shown.

HEK293 cells (ATCC, CRL-1573, lot: 3449904) were cultured in DMEM supplemented with 10% FBS at 37 °C in a humidified incubator containing 5% CO₂. For transfections, cells were suspended and transfected using Lipofectamine 2000 with the appropriate expression constructs as previously described⁴⁹. The cDNA amounts used per well in 96-well plate were as following unless indicated in figure legends: rat GABA_B receptor, GB1 (3 ng, 6 ng, 9 ng, 30 ng) and GB2 (4 ng, 8 ng, 12 ng, 40 ng), mouse α_2AAR (10 ng, 15 ng, 20 ng, 50 ng), mouse CB1 receptor (80 ng), Gα_i1^Nluc (1 ng), Gα_i2^Nluc (3 ng), Gα_i3^Nluc (3 ng), Gα_oA^Nluc (1 ng), Gα_oB^Nluc (1 ng), Gα_z^Nluc (1.5 ng), Gβ₁ (10 ng) and ^VenusGγ₂ (10 ng). The ratio of DNA to Lipofectamine 2000 ratio was 1:2. After a 24 h culture in 96-well plates, the cells were ready for the bioluminescence resonance energy transfer (BRET) experiments.

For the primary culture of neurons, all experiments were specifically designed to minimize the number of animals used and were approved by the Animal Experimentation Ethics Committee of the College of Life Science and Technology, Huazhong University of Science and Technology, Wuhan, China. Kunming mice were obtained from the Center for Disease Control and Prevention of Hubei Province. The mice were raised in a specific pathogen free (SPF) environment with an ambient temperature of 18–22 °C, a humidity of 50%-60%, and a 12 h light-dark cycle.

Primary cerebellar granule neuronal cultures were prepared from one-week-old newborn Kunming mice as previously described³⁹. The dissected tissue was gently triturated after Versene (15040066; Gibco, Shanghai, China) treatment for 5 min at 37 °C, and the homogenate was centrifuged at 170 g for 5 min. In the meantime, mixtures of the DNA and Lipofectamine 2000 (Ref 11668019; Thermo Fisher Scientific, Shanghai, China) in a 1:3 ratio in an Opti-minimal essential medium (Opti-MEM) (Ref 31985070; Thermo Fisher Scientific, Shanghai, China) were prepared following the manufacturer’s protocols and as previously described⁴². The pellet was re-suspended in culture medium DMEM-F12 (Ref 11320-033; Gibco) supplemented with 2 mM glutamine, 30 mM KCl, 100 U/mL penicillin, 100 μg/mL streptomycin, and 10% foetal bovine serum (FBS) and seeded into 96-well plates (100 μL/well) previously coated with poly-L-ornithine (Sigma-Aldrich). Then, the mixtures of DNA and Lipofectamine 2000 (50 μL/well) were added to the wells. BRET experiments were performed three or four days after transfection. BRET experiments were performed at DIV3 or DIV4. The cDNA amounts used per well in 96-well plate were as following: Gα_i1^Nluc (25 ng), Gα_i2^Nluc (75 ng), Gα_i3^Nluc (75 ng), Gα_oA^Nluc (25 ng), Gα_oB^Nluc (25 ng), Gα_z^Nluc (25 ng), ^VenusGγ₂ (50 ng), ^VenusGγ₈ (50 ng) and ^VenusGγ₉ (50 ng).

Primary cortical and hippocampal neuronal cultures were prepared from embryonic day 17.5 mice as previously described⁸⁵. The cortex or hippocampi were digested with trypsin, and cells were seeded in neurobasal medium (Ref 21103049; Gibco) supplemented with 2% B27 (Ref 17504-044; Gibco), 4 mM GlutaMAX, 25 μM glutamic acid, 100 U/mL penicillin, 100 μg/mL streptomycin and 10% FBS in 96-well plates. After three days in culture (DIV3), the culture medium was supplemented with 5 μM cytosine β-D-arabinofuranoside hydrochloride (C6645; Sigma-Aldrich) and incubated overnight. Then, 75% of the medium was replaced by neurobasal medium supplemented with B27, GlutaMAX and antibiotics. Neurons were then transfected with Gα_i1^Nluc or Gα_oA^Nluc (50 ng/well), together with ^VenusGγ₂ (50 ng/well) using Lipofectamine 2000 on DIV7. The ratio of DNA to Lipofectamine 2000 ratio was 1:2. BRET experiments were performed on DIV10 or DIV11.

Cerebellar granule neurons (CGNs) and transfected HEK293 cells were fixed with 4% formaldehyde and blocked with 2% bovine serum albumin (BSA) and 0.1% Triton X-100 in phosphate-buffered saline (PBS). The CGNs were incubated with primary GFP antibody (1:200; ab1218, Abcam, Shanghai, China) at 4 °C overnight. After washing three times with PBS, the cells were incubated with secondary anti-mouse antibody Alexa Fluor® 488 AffiniPure Donkey Anti-Mouse IgG (H + L) (1:500, 715-545-150, Jackson ImmunoResearch, Shanghai, China) at 25 °C for 2 h. The CGNs were then stained with DAPI for 15 min. The cells were washed with PBS and mounted with FluorSave reagent (AR1109, Boster Biological Technology Co. Ltd., Wuhan, China). Images were obtained with an Olympus FV1000 laser scanning confocal microscope (60 x objective for HEK293 cells and 40x objective for CGNs, Olympus Corporation, Tokyo, Japan) equipped with appropriate fluorescence and filters (FITC: 488/530 nm; DAPI: 405/449 nm). The images were digitized and saved in TIFF format.

CGNs were starved in HEPES-buffered saline (HBS) containing 10 mM HEPES pH 7.4, 140 mM NaCl, 4 mM KCl, 2 mM MgSO₄, and 1 mM KH₂PO₄. Cortical and hippocampal neurons were starved with artificial cerebrospinal fluid (aCSF) buffer containing 140 mM NaCl, 2 mM CaCl₂, 3 mM KCl, 10 mM HEPES, and 10 mM D-glucose, at 37 °C for 1 h before the BRET experiments. BRET measurements were performed as previously described using the Mithras LB 940 multimode microplate reader (Berthold Technologies, Bad Wildbad, Germany)⁴⁹ with the programme MikroWin (Version 4.41) or PHERAstar FS (BMG Labtech, USA)⁸⁶ with the programme PHERAstar control (Version 4.00 R4). The signals emitted by the donor (460–500 nm band-pass filter, Em 480 nm) and the acceptor (510–550 nm band-pass filter, Em 530 nm) were recorded by Mithras LB 940 after the addition of 10 μM furimazine. All measurements were performed at 37 °C. The BRET signal was determined by calculating the ratio between the emission of acceptor and donor (Em 530 nm / Em 480 nm). The basal BRET ratio (BRET_basal) of cells was recorded before the stimulation with drugs or buffer. The change in BRET ratio (net BRET) was obtained by subtracting the BRET ratio after agonist treatment from the basal BRET (BRET_basal-BRET_agonist). Agonist-induced change in BRET ratio for the different G_i/o sensors was expressed as percentage of the basal signal ((BRET_basal-BRET_agonist/BRET_basal) x100). To study the kinetics, the BRET was measured in real time with a counting time of 0.5 s, and the drugs were injected using the Mithras LB 940 injectors at the indicated time.

Total RNA was extracted using standard methods (Trizol, Invitrogen) from CGNs and HEK293 cells. Quantitative reverse transcription PCR (qRT-PCR) was carried out with SYBR Green (Vazyme Biotechnology, Nanjing, China) according to the manufacturer’s protocol as reported⁸⁷. Rplp0 was used as an internal reference for normalization, and the ΔΔCt method was adopted to analyse quantitative PCR data. The primer sets used were referred to previous references^75,88. For mouse (CGNs), Gnai1 (NM_010305), Forward (Fw): 5’-AAGCTGACTCGCCTTCCCAG-3’, Reverse (Rv): 5’-GTAGTTTACAGTTCTCCACACG-3’; Gnai2 (NM_008138), Fw: 5’-TGCCTTGAGTGTGTCTGCGTG-3’, Rv: 5’-CTCAGTGACGTTGGCAGTTG-3’; Gnai3 (NM_010306), Fw: 5’-GTGCAGTCCGTGTACAAGAG-3’, Rv: 5’-GATGAATGGATCCGAGCCAC-3’; Gnao1 transcript isoform A (NM_010308) Fw: 5’-AGGAAGACGGACTCCAAGATG-3’, Rv: 5’-AGTCGAAGAGCATGAGAGAC-3’; Gnao1 transcript isoform B (NM_001113384) Fw: 5’-AGGAAGACGGACTCCAAGATG-3’, Rv: 5’-AGATGTGTCTGTGAACCACTTG-3’; Gnaz (NM_010311), Fw: 5’-CAGCCGTGCTTAGAAACATCG-3’, Rv: 5’-TCTAGTGACACTCCACCTCC-3’ and Rplp0 (NM_007475), Fw: 5’-CTCACTGAGATTCGGGATATG-3’, Rv: 5’-CTCCCACCTTGTCTCCAGTC-3’. For human (HEK293 cells), GNAI1 (NM_002069), Fw: 5’-CATCTCTGACCTTGTTTCAGC-3’, Rv: 5’-CTTCAACCCAGTGACAACACG-3’; GNAI2 (NM_002070), Fw: 5’-ACTCCGTGCCTTGAGTGTG-3’, Rv:5’-TTGTCTGGAACAGCCCTTGG-3’; GNAI3 (NM_010306), Fw: 5’-GGAAAGTTACGTTCACTTCAACC-3’, Rv: 5’-TTGGACCCCAAAAGGCACTG-3’; GNAO1 transcript isoform A (NM_020988) Fw: 5’-AGAAAGGCTGACGCCAAGAT-3’, Rv: 5’-AGTCGAAGAGCATGAGAGAC-3’; GNAO1 transcript isoform B (NM_138736) Fw: 5’-AGAAAGGCTGACGCCAAGAT-3’, Rv: 5’-TGGACGTGTCTGTGAACCAT-3’; GNAZ (NM_002073), Fw: 5’-CTACGAGGATAACCAGAC-3’, Rv: 5’-TACGTGTTCTGGCCCTTG-3’ and RPLP0 (NM_053275), Fw: 5’-ATGCAGCAGATCCGCATGT-3’, Rv: 5’-TTGCGCATCATGGTGTTCTT-3’.

The cell membranes were prepared as reported⁸⁹. HEK293 cells or CGNs were washed three times with PBS, then scraped and collected by centrifugation at 170 g for 5 min. Cells were resuspended in 500 μl Tris buffer (50 mM Tris pH 7.4, 50 mM NaCl) with cOmplete protease inhibitor cocktail (Roche) and crushed through a 26 gauge 5/8 inch needle attached to a syringe for 30 passages. After centrifugation at 860 g at 4 °C for 5 min, liquid supernatants were transferred to high speed centrifuge tubes and centrifuged at 44000 g at 4 °C for 20 min by Optima MAX-TL ultracentrifuge (Beckman Coulter, Brea, CA, USA). The precipitated membranes were diluted gently in 50 μl Tris-NaCl buffer. The protein amount was further determined using BCA protein assay kit and 20 μg protein was loaded for western blotting detection.

Cells were treated using RIPA lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, sodium orthovanadate, sodium fluoride, EDTA, leupeptin; Beyotime Bio., Cat. P0013C, China) and protein concentrations were determined using the BCA protein assay kit. Equal amounts of protein (20 μg) from total cell lysis or cell membranes were separated by SDS–polyacrylamide gel electrophoresis (PAGE) on 10 to 12% gels. Proteins were transferred to nitrocellulose membranes and washed in blocking buffer (5% nonfat dry milk in Tris-buffered saline and 0.1% Tween 20) for 2 h at 25 °C. The blots were incubated with the primary monoclonal antibodies anti-GB1 mAb (1:1000, ab55051, Abcam, Shanghai, China), anti-Gβ₂ rabbit (1:1000, A9643, ABclonal Technology, Wuhan, China), anti-β-actin (1:3000, KM9001T, Sungene Biotech., Tianjin, China), the rabbit polyclonal antibodies anti-CB1 (1:1000, A1447, ABclonal, Wuhan, China), anti-α_2AAR (1:1000, A2809, ABclonal, Wuhan, China), anti-Gβ₁ (1:1000, A1867, ABclonal, Wuhan, China), anti-Gβ₃ (1:1000, A1387, ABclonal, Wuhan, China) and anti-Gβ₅ (1:1000, A4447, ABclonal, Wuhan, China). The primary antibodies were at the relevant dilution overnight at 4 °C and then incubated with DyLight 800 4 X PEG conjugated secondary antibodies (1:20,000, anti-mouse IgG, #5257; 1:20,000, anti-rabbit IgG, #5151, Cell Signaling Technology, Shanghai, China) for 2 h at 25 °C. Membranes were imaged using an Odyssey infrared scanner (LI-COR Biosciences, Lincoln, NE, USA) at 700 nm.

Results are presented as the mean ± SEM of at least three independent experiments. Statistical analysis was performed using GraphPad Prism 9.5.1 software (GraphPad Software Inc., San Diego, CA, USA). Dose-response experiments were analysed using nonlinear curve fitting for the log (agonist) vs. response (three parameters) curves. Statistical analysis was performed using the Ordinary one-way ANOVA with a Dunnett’s post-hoc multiple comparison test or unpaired t-test (two-tailed) or paired t-test. P < 0.05 was considered to be statistically significant.

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary Information

Source Data