Specific pharmacological and Gi/o protein responses of some native GPCRs in neurons

G protein-coupled receptors (GPCRs) constitute the largest family of membrane proteins and are important drug targets. The discovery of drugs targeting these receptors and their G protein signaling properties are based on assays mainly performed with modified receptors expressed in heterologous cells. However, GPCR responses may differ in their native environment. Here, by using highly sensitive Gi/o sensors, we reveal specific properties of Gi/o protein-mediated responses triggered by GABAB, α2 adrenergic and cannabinoid CB1 receptors in primary neurons, different from those in heterologous cells. These include different profiles in the Gi/o protein subtypes-mediated responses, and differences in the potencies of some ligands even at similar receptor expression levels. Altogether, our results show the importance of using biosensors compatible with primary cells for evaluating the activities of endogenous GPCRs in their native environment.

G protein-coupled receptors (GPCRs) are ubiquitously expressed in every cell type, and constitute the largest family of membrane proteins 1 .They participate in the regulation of a large variety of physiological functions 2 .Their dysregulation or malfunction can be the cause of numerous diseases 3 .Accordingly, controlling their activity with selective drugs can have multiple therapeutic effects, as illustrated by the large proportion of the clinical drugs targeting GPCRs 4 .
Most of the physiological functions of GPCRs are mediated through the coupling of G proteins 2 .Eventually, GPCRs are able to activate several G protein subtypes, such as G i/o , G s , G q , and G 12/13 family [5][6][7][8] , but this may vary depending on the use of ligands, a phenomenon called ligand biased effect 9,10 , or/and the cellular environment known as system bias 11 .This concept of functional selectivity 11 , that combined effect of ligand and system bias, has been largely used in the characterization of ligands [12][13][14] , but the signalling properties and identification of potential drug candidates targeting GPCRs are performed in heterologous cells.Since the cellular environment is critical, the analysis of G protein response profile of endogenous receptors in their native environment is therefore of much interest to validate drug candidates.Indeed, it becomes clear that many interacting proteins, post-translational modifications, localization in specific compartments, lipid membrane and ion environments largely influence GPCR signalling properties 15,16 .It is thus essential to examine the effects of drugs on the signalling properties of GPCRs in their native environments.
Functional analysis of endogenous GPCRs in their native environment is usually performed through the measurement of second messengers, such as cAMP, inositol phosphate or Ca 2+17 , or ion channel regulation 2,18 .Due to receptor reserves and signal amplification, such assays are not precise for estimating drug activity, potency and efficacy 9 .G proteins are first effectors of GPCRs, such that measuring their activation is less prone to signal amplification 18 .GTPγS binding assay can detect G protein activation in native cells or tissues 19,20 .However it requires the preparation of membranes and these assays are not specific for one G protein subtype, unless an immunoprecipitation of the G protein is added 21 .Nowadays, much hope in studying native GPCRs is based on the development of specific optical biosensors that must be sensitive enough to detect the activity of native receptors 22 usually expressed at very low levels, but most of them are still not compatible with these endogenously expressed receptors 13,[23][24][25][26] .
In the central nervous system, GPCRs are essential regulators of synaptic transmission, acting at both pre-and post-synaptic levels, as well as in glial cells 27 .More than half of neuronal GPCRs can couple to one or more of the various G i/o protein subtypes 5 , but their expression profiles vary in different native tissues or cell types and specific cellular function depends on distinct G i/o protein subtypes 12,28,29 .It is therefore essential to determine which G protein can be activated by a GPCR upon stimulation with various agonists in its native environment.
In the present study, we describe BRET-based sensors for each G i/o protein subtypes that can be used to study endogenous GPCRs in living neurons.Our study focuses on three types of GPCRs that play important roles in modulating synaptic activity, the GABA B 30 , α 2 adrenergic 31,32 and cannabinoid CB1 receptors 33 .We could detect the activation of G i/o proteins from a small number of neurons and monitor both the kinetics and dose-response of the effect mediated by various compounds including agonists, antagonists and positive allosteric modulators, in different types of neurons.Our data reveal differences in the profile of G i/o protein subtypes mediated responses and agonist potencies between these recombinant and native receptors in HEK293 cells and neurons, respectively.In addition, our results show a major difference in G i1 versus G oA protein activation induced by a CB1 receptor agonist in neurons, but not in HEK293 cells.Finally, different composition of Gγ subunits in the neurons can also lead to specific G i/o protein responses.Altogether, our results reveal the importance of evaluating GPCR activities in their native environment and highlight the need for sensitive biosensors compatible with native receptors in their natural environment.

G i/o protein sensors for endogenous GPCRs in neurons
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][24][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 ( Venus Gγ), 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 Venus Gγ.We used mostly the mouse cerebellum granule neurons (CGNs) that constitute the largest homogenous neuronal population of the mammalian brain 34 .
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 Nluc 36 , 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 i3 37 .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 Gγ 2 subunit with the amino acid sequence KLGT serving as a linker, and the construct is named Venus Gγ 2 (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
).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 Venus Gγ 2 (Supplementary Fig. 1c).In addition, the basal BRET ratio of the constructs Gα i1 Nluc and Gα oA Nluc co-transfected with Venus Gγ 2 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 coexpression of the constructs Gα Nluc + Venus Gγ 2 for the different Gα subunits did not change the expression level of endogenous Gβ subunits (Supplementary Fig. 1e).Altogether, the G protein Nlucbiosensors are expected to produce a much improved signal-tonoise 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 neurons 38 , including in CGNs 39 .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 + Venus Gγ 2 (Fig. 1d, Supplementary Fig. 2a) and Gα oA Nluc + Venus Gγ 2 (Supplementary Fig. 1d, 2b), while the change was barely detectable with Gα i1 Rluc8 + Venus Gγ 2 (Fig. 1d) and Gα oA Rluc8 + Venus Gγ 2 (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 Venus Gγ 2 or with Venus (Supplementary Fig. 2d).Similarly, no change of BRET signal was measured with the dominant negative mutants Gα i1 Nluc -S47A 40 or Gα i1 Nluc -G202T 41 (Supplementary Fig. 2e) when they are co-expressed with Venus Gγ 2 .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 Gγ 2 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 Venus G γ2 .This low level of transfection is consistent with previous data reported with CGNs 42 , 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 pEC 50 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.

Lower agonist potencies for the endogenous GABA B receptor in neurons
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 tissues 21 .pEC 50 s 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 pEC 50 was similar to that observed with higher expression levels, and still significantly different from that measured in CGNs (Fig. 3g).
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 neurons 43 and have a greater potential for drug development with less side effect 44 , including for the GABA B receptor 45 .It was recently demonstrated that these PAMs bind at the interface of the transmembrane domains of this dimeric receptor [46][47][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 activity 50 , in the absence of agonist, of the GABA B receptor that is usually observed in recombinant systems 47,49,[51][52][53][54][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 reported 49 .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.

Various efficacies of endogenous G i/o -coupled GPCRs in neurons
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 cerebellum 33 compared to CB2 that is mainly expressed at the periphery 57 .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 α 2 adrenoceptor 58 , D 2/4 receptor 59 , 5-hydroxytryptamine receptor 1B/1D 60 and the M 2/4 muscarinic receptors 61 , respectively.In contrast, agonists of the plateletactivating 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 investigated 62 .
Next, using α 2 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 2A AR) (Fig. 5b).It was not due to the high level of expression of α 2A AR, since no discrimination between brimonidine and other agonists was observed in HEK293 cells that expressed α 2A AR at a similar level to that found in CGNs (Fig. 5c).In these conditions (15 ng of α 2A AR cDNA/well), the brimonidine pEC 50 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 2A AR in HEK293 cells and the endogenous one in 0.00  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).
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.

Difference between G i and G o proteins revealed for the endogenous CB1 receptor
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 2A AR (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 α 2 AR, 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).

Major influence of the Gγ subunit on the G protein coupling in CGNs
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 Venus Gγ 2 , Venus Gγ 8 or Venus Gγ 9 , which are well-used in the previously reported G protein BRET sensors 25,36 .Venus Gγ 8 or Venus Gγ 9 belong to two different groups of Gγ subunits (Supplementary Fig. 8).Gγ 8 , like Gγ 2 , is well expressed in some regions of the brain, in contrast to Gγ 9 that seems restricted to retina 66 .Interestingly, a major difference in G protein response profile of the GABA B receptor and α 2 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.

Discussion
Our study compares the pharmacological and G protein responses of native, versus recombinant G i/o -coupled receptors in primary neurons and in heterologous cells, respectively.Such an analysis was made possible thanks to the sensitive BRET-based G i/o protein sensors.These sensors allow a simple analysis of different types of GPCR ligands on each G i/o protein subtypes in primary neurons in a medium throughput format (in 96-well plate).For the same G protein sensor, our study reveals important differences in agonist potencies between neurons and heterologous HEK293 cells (Fig. 8a).GABA B and α 2 adrenergic receptor agonists displayed higher potencies in cell lines versus neurons even at similar receptor expression levels, in contrast to CB1 agonist.And when comparing the profile of response of different G protein sensors in the same cellular system, major differences of potencies can be observed, such as for a CB1 agonist (CP 55940) that shows high potency in mediating G i1 response compared to G oA in neurons.These data highlight the importance of evaluating the activities of endogenous GPCRs in their native environment, and of developing biosensors allowing such analyses.
In order to be able to compare the properties of recombinant and native G i/o -coupled GPCRs, we needed sensors for the various G i/o proteins compatible with measurements of their activity in primary neurons.Here we showed that Nluc-based BRET sensors, could do so, most likely due to the high emission of Nluc, compared to Rluc8 used previously 24,25 , since the efficacy of transfection of these constructs in primary neurons remains very low (around 2%).The principle of our optical biosensors is not novel, since the rearrangement between Gα and Gβγ upon receptor activation using resonance energy transfer techniques was previously reported by FRET 23 , BRET 24,25 including with the use of Nluc 36 .However, these biosensors have never been tested in primary neurons.Interestingly, other kind of biosensors compatible with the detection of endogenous G protein activity were recently reported 22,36 .Among them, the unimolecular "BERKY" sensors were used in live primary cells but they required the use of a viral vector.In addition, these sensors were not able to discriminate between different G protein subtypes, such as between G i1 , G i2 , G i3, G oA, G oB and G z since it is a unimolecular sensor using a peptide recognizing the GTP form of these different G proteins.Accordingly, they cannot be used to analyse the influence of the different Gβγ subunits as our sensors and the previous BRET sensors 24,25,36 do.Instead, our sensors nicely revealed specific responses of each G i/o protein triggered by some native GPCRs in primary neurons, which are different from what can be observed in heterologous cells.They work well for the most abundant neuronal GPCRs, but would likely need to be further improved for receptors expressed at lower levels.We have to be cautious when comparing the profile of response of different G protein sensors in the same cellular system (Figs.7 and 8b).Indeed, each G protein BRET sensor might have its own properties, with its specific range of signal, even though they were constructed similarly, and Gα i/o subtypes have a high sequence identity 37 .G protein sensor response can result either from a complete dissociation between Gα and Gβγ but also from a repositioning of Gβγ relative to Gα without a real dissociation 23 .Within the G i/o protein family, using similar FRET-based G protein sensors between Gα and Gβγ, Frank et al. have proposed that receptor-activated G i1 , G i2 , G i3 and G z undergo subunit rearrangement rather than subunit dissociation, whereas G o proteins either rearrange with a very distinct pattern or dissociate during activation 67 .But even though there is good evidence that the activated Gα and Gβγ subunits actually loose affinity to each other and exchange faster depending of the G proteins, there is no evidence that the subunits separate completely in intact cells 68 .When analysing the same G protein sensor, our study reveals major differences in G protein responses and agonist potencies between neurons and heterologous HEK293 cells.The difference in the profiles of G protein response using different Gγ subunits (Gγ 2 , Gγ 8 or Gγ 9 ) (Fig. 7) shows clearly the impact of Gγ in CGNs.It shows that not only the Gα subunit is important, but also the Gβ and Gγ, consistent with other recent studies 65,66 .Therefore, it brings to our attention that the G protein-mediated response also depends detected on the biosensor components used, which is the limitation of biosensors that only detect the response of a well-defined G protein.Meanwhile, the difference can be due to the intrinsic properties of the system under study.For example, the association of these receptors with specific protein partners that may influence their coupling to some Gα i subtypes, present in neurons but not in HEK293 cells.Indeed, specific proteins interacting with the GABA B receptor were identified in brain tissues 15,69 , while they are not expressed in HEK293 cells.For example, the soluble form of APP, AJAP-1 and PIANP bind to the extracellular sushi domain of GABA B 70,71   , TRPV1 channels interact in the membrane 72 , and the intracellular KCTDs and 14-3-3 proteins 73 modulate GABA B receptor downstream signalling.The KCTD proteins are constitutively associated with GABA B and they control its kinetics of activation 73 .
Our sensors showed a difference between Gi and Go proteins revealed by a CB1 agonist CP 55940, which has a higher potency in generating G i1 than G oA responses in neurons, relative to the other agonists tested.Indeed, such biased signalling between G i and G o proteins have already been reported upon agonist stimulation 12,13,74 , by allosteric modulators 49 or as a result of genetic variation 75 .In our study, the molecular bases of this difference in agonist effect is unclear.CP 55940 and Win 55212-2 (K i 1.1 nM and 62.3 nM for CB1 receptor, respectively) 76,77 have different scaffolds.Their differential effects might be explained by the ligand-binding kinetics 78 , by specific active conformation stabilized by the agonist 79 , or due to specific components associated to the receptor in these neurons, better stabilizing the CP 55940/CB1/Gi complex.Our sensors also revealed larger differences in some agonist potencies in neurons compared to transfected HEK293 cells.This is nicely illustrated for the CB1 and GABA B agonists CP 55940 and APPA, respectively, that show a higher potency than the other agonists in neurons (Fig. 8a).In addition and unexpectedly, no intrinsic agonist activity of the PAM Rac BHFF was detected with the native GABA B receptor in contrast to the recombinant one in HEK293 cells.Finally, the important differences in GABA B receptor agonist potencies that are higher in transfected HEK293 cells compared to neurons are most probably not due to the higher expression in the transfected cells in our study.Indeed, a 10-times higher affinity for GABA in neurons compared to HEK cells 80 and similar agonist potencies in GTPγS binding experiments in the two systems 81 were previously reported.Possible explanations might come from the environment of the receptor in neurons, such as its compartmentalization and its interaction with extracellular and intracellular proteins 15,50 , including other endogenous receptors.
Our sensors also revealed specific ligand-independent G protein response in neurons compared to HEK293 cells.Indeed, no detectable constitutive activity of the GABA B receptor was observed in CGNs in contrast to transfected cell lines 49,51,52 .As discussed above, this lack of constitutive activity might be explained by the environment of the receptor in neurons.The importance of the intracellular protein partners for the constitutive activity of a GPCR was nicely illustrated for the postsynaptic metabotropic glutamate receptor mGlu5.Its constitutive activity was maintained low in neurons by the long isoform of Homer, while it is revealed when the short isoform of Homer is expressed 82 .
In conclusion, our study reveals the importance of evaluating the pharmacological and G protein response properties of native GPCRs in Agonist-induced change in BRET ratio (% of basal) Agonist-induced change in BRET ratio (% of basal)  and a, b) normalized to the G i/o sensor that has the highest response: for Gγ 2 , G i1 for both the GABA B and CB1, and Gα z for α 2 AR; for Gγ 8 , G oA for GABA B and α 2 AR and G i1 for CB1; and for Gγ 9 , G i1 for GABA B and CB1, and Gα z for α 2 AR.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.
primary cells to better understand their signalling, identify and characterize ligands expected to have therapeutic effects.It also reveals the powerfulness of using G protein sensors compatible with native cells, though biosensors compatible with single cell analysis and imaging will certainly help revealing the potential cell-to-cell or sub-compartment heterogeneity in GPCR and G protein coupling 83,84
Venus was inserted at the N-terminus of Gγ 2 , Gγ 8 and Gγ 9 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 cell culture and transfection
HEK293 cells (ATCC, CRL-1573, lot: 3449904) were cultured in DMEM supplemented with 10% FBS at 37 °C in a humidified incubator containing 5% CO 2 .For transfections, cells were suspended and transfected using Lipofectamine 2000 with the appropriate expression constructs as previously described 49   (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.

Primary neuron culture and transfection
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  Fig. 8 | Major differences in agonist potencies and the G protein response profile for some GPCRs between heterologous and native cells.a Scheme illustrating the difference in agonist or PAM potencies for the GABA B and CB1 receptors between HEK293 cells and CGNs for the G i1 sensor.For each cell and receptor, the relative size of the triangles is according to the pEC 50 values of the indicated agonists.b Scheme illustrating the difference in G i/o protein response (G i1 , G i2 , G i3 , G oA , G oB and G z ) between CGNs and HEK293 cells for the receptors GABA B , CB1 and α 2 AR in presence of the indicated agonists, when Gγ 2 is used for all these G i/o sensors.For each cell and receptor, the size of the circle is the percentage of change in BRET ratio upon the agonist stimulation for GABA B (Fig. 2a), α 2 AR (Fig. 5e) and CB1 (Fig. 6b) normalized to the G i/o sensor that has the highest response: in HEK293 cells, G oA for GABA B , α 2 AR and G i1 for CB1, and in CGNs, G i1 for both the GABA B and CB1, and G z for α 2 AR.
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 described 39 .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 described 42 .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 ( Primary cortical and hippocampal neuronal cultures were prepared from embryonic day 17.5 mice as previously described 85 .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 Venus Gγ 2 (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.

BRET measurement
CGNs were starved in HEPES-buffered saline (HBS) containing 10 mM HEPES pH 7.4, 140 mM NaCl, 4 mM KCl, 2 mM MgSO 4 , and 1 mM KH 2 PO 4 .Cortical and hippocampal neurons were starved with artificial cerebrospinal fluid (aCSF) buffer containing 140 mM NaCl, 2 mM CaCl 2 , 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.

Quantitative reverse transcription PCR
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 reported 87 .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'-AAGCT-buffer.The protein amount was further determined using BCA protein assay kit and 20 μg protein was loaded for western blotting detection.

Statistical analysis
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).Doseresponse 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 (twotailed) or paired t-test.P < 0.05 was considered to be statistically significant.

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

Fig. 1 |
Fig. 1 | Design and validation of the G i/o protein sensors in neurons.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 Venus Gγ 2 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 Venus Gγ 2 .d Kinetics of the BRET signal between the indicated Gα constructs (Gα i1 Nluc or Gα i1 Rluc8) and Venus Gγ 2 in CGNs.Buffer or baclofen (100 μM) were injected at the indicated time (arrow).The data are representative of BRET ratios

Fig. 3 |
Fig. 3 | Lower agonist potencies of the endogenous GABA B receptor in CGNs. 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 (EC 80 ) 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 (pEC 50 ) between HEK293 cells and CGNs determined by the G i1 (e) or G oA (f) BRET sensors.Dotted line is the correlation of pEC 50 determined with HEK293 cells.Red line is fit of pEC 50 in CGNs with the same slope as the dotted line.g The pEC 50 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 pvalues are available in source data provided as a Source Data file.

15 C
id in e X y la z in e O x y m e t a z o li n e T iz a n id in e B r im o n id in e C lo n id in e X y la z in e O x y m e t a z o li n e T iz a n id in e B r im o n id in e C lo n id in e X y la z in e O x y m e t a z o li n e T iz a n id in e B lo n id in e X y la z in e O x y m e t a z o li n e T iz a n id in e B r im o n id in e C lo n id in e X y la z in e O x y m e t a z o li n e T iz a n id in e B

Fig. 5 |
Fig. 5 | Activity of endogenous α 2 AR detected by the G i/o protein sensors.a Effect of the indicated α 2 AR agonists on the net BRET of the G i1 and G oA sensors in CGNs co-transfected with Gα i1 Nluc and Venus Gγ 2 or Gα oA Nluc and Venus Gγ 2 (amounts of cDNA as in Fig. 2a).b Effect of the indicated α 2 AR agonists on the net BRET of the G i1 and G oA sensors in HEK293 cells co-transfected with the mouse α 2A AR (10 ng, 20 ng or 50 ng), and Gβ 1, Venus Gγ 2 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 α 2 AR in cell membranes of CGNs, mock-transfected HEK293 cells and HEK293 cells transfected with indicated amount of α 2 AR, by western blotting.Values are mean ± SEM normalized as fold of α 2 AR expression in CGNs from four independent experiments.d The pEC 50 of brimonidine in CGNs (n = 3) and transfected HEK293 cells (n = 4) with indicated amount of α 2 AR measured by G i1 or G oA sensor.e Percentage of change in BRET ratio between Gα Nluc and Venus Gγ 2 induced by brimonidine between Gα Nluc and Venus Gγ 2 in HEK293 cells (n = 3) (α 2 AR: 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.

Fig. 6 |
Fig. 6 | Difference of G i1 and G oA responses by cannabinoid receptor CB1 agonists.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 96well 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 Venus Gγ 2 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 Venus Gγ 2 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 (pEC 50 ) between HEK293 cells and CGNs determined by the G protein sensors.Dotted lines are the correlation of pEC 50 determined with HEK293 cells.Red lines are the fit of pEC 50 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.

Fig. 7 |
Fig. 7 | Gγ subunit influences G protein responses in CGNs.a, b Percentage of change in BRET ratio between Gα Nluc and Venus Gγ 8 (a) or Venus Gγ 9 (b) induced by baclofen, brimonidine and CP 55940 in CGNs for the indicated G i/o proteins.CGNswere co-transfected with the indicated Venus Gγ 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 (Gγ 2 , Gγ 8 and Gγ 9 ) responses in CGNs for the receptors GABA B , CB1 and α 2 AR 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), α 2 AR (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 Gγ 2 , G i1 for both the GABA B and CB1, and Gα z for α 2 AR; for Gγ 8 , G oA for GABA B and α 2 AR and G i1 for CB1; and for Gγ 9 , G i1 for GABA B and CB1, and Gα z for α 2 AR.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.
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: VenusGγ 9 (50 ng).