Complementary biosensors reveal different G-protein signaling modes triggered by GPCRs and non-receptor activators

It has become evident that activation of heterotrimeric G-proteins by cytoplasmic proteins that are not G-protein-coupled receptors (GPCRs) plays a role in physiology and disease. Despite sharing the same biochemical guanine nucleotide exchange factor (GEF) activity as GPCRs in vitro, the mechanisms by which these cytoplasmic proteins trigger G-protein-dependent signaling in cells have not been elucidated. Heterotrimeric G-proteins can give rise to two active signaling species, Gα-GTP and dissociated Gβγ, with different downstream effectors, but how non-receptor GEFs affect the levels of these two species in cells is not known. Here, a systematic comparison of GPCRs and three unrelated non-receptor proteins with GEF activity in vitro (GIV/Girdin, AGS1/Dexras1, and Ric-8A) revealed high divergence in their contribution to generating Gα-GTP and free Gβγ in cells directly measured with live-cell biosensors. These findings demonstrate fundamental differences in how receptor and non-receptor G-protein activators promote signaling in cells despite sharing similar biochemical activities in vitro.


Introduction
Heterotrimeric G-proteins are ubiquitous molecular switches that transduce extracellular signals into intracellular cascades of biochemical reactions to steer cellular responses (Gilman, 1987). The ON/ OFF state of these switches is defined by their nucleotide binding status-GDP-bound G-proteins are OFF, whereas GTP-bound G-proteins are ON. The switching between states is primarily determined by two biochemical events. The first one is nucleotide exchange of GDP for GTP, which determines the rate of activation, and the second one is hydrolysis of GTP to GDP, which determines the rate of inactivation. While these two reactions can be carried out spontaneously by G-proteins, they are tightly regulated by the enzymatic activity of other proteins in the cellular context. The G-protein regulatory mechanism controlled by G-protein-coupled receptors (GPCRs) is the best characterized to date (Pierce et al., 2002;Weis and Kobilka, 2018). GPCRs in their active conformation are guanine nucleotide exchange factors (GEFs) that catalyze nucleotide exchange on the Ga subunit of Ga-Gbg heterotrimers. Upon GTP loading, Gbg disengages from Ga, leading to the formation of two active signaling species, that is, Ga-GTP and free Gbg, that can engage their respective intracellular effectors to initiate signaling cascades. In addition to GPCR-mediated activation, G-proteins are regulated by many cytoplasmic proteins with different enzymatic activities (Sato et al., 2006;Siderovski and Willard, 2005). These include proteins that accelerate the intrinsic rate of nucleotide hydrolysis by G-proteins (GTPase accelerating proteins [GAPs]) (Ross and Wilkie, 2000), proteins that block nucleotide exchange (guanine nucleotide dissociation inhibitors [GDIs]) (Blumer et al., 2012), or even GEFs that are not GPCRs (Cismowski et al., 2000;DiGiacomo et al., 2018;Tall, 2013; Figure 1A).
From a historical perspective, the heterotrimeric G-protein research field has relied heavily on reductionist approaches. Breaking down complex systems into defined biochemical entities that can be reconstituted and rigorously characterized in vitro has been a powerful approach to subsequently understand the mechanisms operating in cells. This approach has proven particularly successful for understanding GPCR-mediated regulation of G-proteins, a mechanism with broad biomedical implications that has been extensively characterized at the molecular level and that represents one of the most widely pursued pharmacological targets (Sriram and Insel, 2018). Similar biochemical reductionist approaches have also been useful to define the enzymatic activities of cytoplasmic regulators of G-proteins like GAPs, GDIs, and non-receptor GEFs, but have left several open questions about how these regulatory mechanisms operate in cells. For example, if G-protein signaling activity in cells is defined by the formation of Ga-GTP or free Gbg, as each one of these species is sufficient to trigger downstream signaling, it is unclear how some GDIs and non-receptor GEFs affect G-protein signaling as a whole. For example, GDIs of the GoLoco motif family have paradoxical effects on G-proteins. On one hand, they bind to inactive, GDP-loaded Ga subunits to prevent the formation of Ga-GTP in vitro (De Vries et al., 2000;Kimple et al., 2002;Natochin et al., 2000). On the other hand, they also prevent the association of Gbg with Ga-GDP (Ghosh et al., 2003;Webb et al., 2005). In fact, GoLoco motif GDIs were originally identified in a genetic screen for 'activators of G-protein signaling (AGS)' in yeast that relied on signaling readouts activated by free Gbg (Cismowski et al., 1999). The situation with several non-receptor GEFs is similarly unclear. A group of non-receptor GEFs characterized by a Ga-binding-and-activating (GBA) motif has been shown not only to promote nucleotide exchange in vitro, but also to physically displace Gbg from GDP-bound Ga Garcia-Marcos et al., 2009;Maziarz et al., 2018), raising the question of approach used to control the activating input for G-protein regulators and simultaneously monitor the G-protein activity output. Chemically induced recruitment of various G-protein regulators to the vicinity of G-proteins at the plasma membrane is achieved through rapamycin-mediated dimerization of FKBP and FRB domains, and G-protein activity is recorded using live-cell bioluminescence resonance energy transfer (BRET) biosensors for free Gbg or Gai-GTP. Gbg tagged with a split YFP (Venus) binds to the C-terminal domain of GRK3 (GRK3ct) fused to nanoluciferase (Nluc) when dissociated from Ga, resulting in BRET from Nluc to YFP. Gai3 internally tagged with a YFP (Citrine) at the ab/ac loop binds to the synthetic sequence KB-1753 fused to Nluc when bound to GTP, resulting in BRET from Nluc to YFP. The online version of this article includes the following source data and figure supplement(s) for figure 1:  what is the relative contribution of each mechanism, Ga-GTP formation and Gbg release, to downstream signaling in cells. Other non-receptor GEFs like Ric-8 proteins promote nucleotide exchange on monomeric Ga but not on Ga-Gbg heterotrimers (Tall et al., 2003), which has led to the contentious speculation that Ric-8 proteins might not regulate directly G-protein activity in cells, but rather work primarily as folding chaperones Tall et al., 2013). Yet another non-receptor GEF, AGS1 (a.k.a. Dexras1), has been shown to activate nucleotide exchange on both monomeric Ga or Ga-Gbg heterotrimers in vitro (Cismowski et al., 2000), but how it influences G-protein signaling in cells is not understood well.
Despite these gaps in mechanistic knowledge, activation of G-proteins by cytoplasmic proteins has been proven to impact various cellular processes and its dysregulation to be linked to different pathologies. This has been made particularly evident for non-receptor GEFs of the GBA family, for which the G-protein regulatory activity can be specifically disabled through mutagenesis (Coleman et al., 2016;de Opakua et al., 2017;Garcia-Marcos et al., 2009;Garcia-Marcos et al., 2012;Maziarz et al., 2018). This surgical approach has been leveraged to establish that G-protein regulation by GBA proteins, like GIV(a.k.a. Girdin) and DAPLE, is involved in normal physiological processes (e.g., formation of the neural tube during embryonic development), or in disease (e.g., cancer metastasis or birth defects) Garcia-Marcos et al., 2015;Ghosh, 2015;Leyme et al., 2017;Marivin et al., 2019). Overall, the involvement of these cytoplasmic regulators in (patho)physiological processes make imperative a more detailed understanding of their mechanisms of action in cells.
The slow progress in understanding G-protein regulation by cytoplasmic proteins compared to their regulation by GPCRs could be due to two experimental issues. One is that the G-protein regulatory function of many GPCRs can be stimulated (and inhibited) with high precision by the simple addition of extracellular ligands, whereas this is not possible for cytoplasmic regulators of G-proteins. The second experimental issue is that approaches to directly detect G-protein activity in cells have typically relied on the detection of Ga-Gbg dissociation instead of detecting Ga-GTP. While these two signaling events correlate well in the process of GPCR-mediated G-protein activation, this is not necessarily the case for many cytoplasmic regulators of G-proteins, making evident the need for detecting both free Gbg and Ga-GTP formation to understand how they operate in cells. Here, a cell-based approach was developed and implemented to overcome current limitations to study the mechanisms of G-protein regulation by cytoplasmic proteins. For this, the action of individual cytoplasmic regulators on their cognate G-proteins was triggered with an exogenous small molecule, and the responses evaluated with real-time biosensors for both Ga-GTP and free Gbg. The resulting experimental system allows to precisely modulate and detect G-protein activity, similar to what can be done biochemically with purified proteins in vitro, but in the more physiologically relevant environment of the cell. This newly developed approach allowed to pinpoint key differences between the modes of G-protein signaling regulation in cells exerted by various proteins with GEF activity, including both receptor (i.e., GPCRs) and non-receptor proteins.

Direct interrogation of G-protein activity regulation by cytoplasmic proteins in cells
To dissect the specific impact of cytoplasmic proteins on the activity of heterotrimeric G-proteins in cells, a strategy to control the signal input and simultaneously assess possible signal outputs was envisioned ( Figure 1B). These studies were focused on Gi proteins because this is the group of heterotrimeric G-proteins for which cytoplasmic regulators have been discovered and characterized more extensively. The premise to establish control over the input is that triggering the relocalization of G-protein regulators from the cytosol to the plasma membrane would allow their action on their constitutively membrane-anchored Gi protein substrates by virtue of increasing the local concentration of the reactants. This was achieved by implementing chemically induced dimerization with rapamycin, which has been successfully applied in the past to rapidly modulate Gi signaling with some GAPs and non-receptor GEFs (Muntean and Martemyanov, 2016;Parag-Sharma et al., 2016). The simultaneous assessment of signaling outputs was carried out by using optical biosensors based on bioluminescence resonance energy transfer (BRET). Because G-protein signaling can be propagated via either Ga-GTP or free Gbg subunits, biosensors for each one of these two active species were implemented in parallel ( Figure 1B). The free Gbg biosensor is based on a previous design by Hollins et al., 2009;Masuho et al., 2015, whereas the Gai-GTP biosensor is based on a recently described design by Maziarz et al., 2020. In both cases, activity is reported as an increase in BRET due to binding of fluorescent protein (FP)-tagged G-protein to a luciferase-tagged protein module that specifically binds to either dissociated Gbg (i.e., the C-terminal region of GRK3, GRK3ct) or GTP-bound Gai (i.e., the synthetic peptide KB-1753).
Three unrelated non-receptor GEFs were investigated: GIV (Garcia-Marcos et al., 2009), AGS1 (Cismowski et al., 2000), and Ric-8A (Tall et al., 2003). For comparison, the GoLoco motif of RGS12 (R12 GL), which has GDI instead of GEF activity in vitro (Kimple et al., 2001), was also investigated ( Figure 1B), and the M4 muscarinic receptor (M4R), a prototypical Gi-activating GPCR, was used as an internal reference to benchmark responses. All cytoplasmic G-protein regulators were fused to the rapamycin-binding domain FKBP separated by a flexible linker. For GIV, Ric-8A and R12 GL, only the specific domains or motifs that are necessary and sufficient to regulate G-protein activity in vitro were used in the constructs (see 'Materials and methods' for details). This was done to avoid potential confounding factors for the interpretation of results, like indirect effects on G-protein signaling or undesired association with membranes in the absence of rapamycin via other domains of the proteins. Along the same lines, the prenylation CAAX motif of AGS1 was mutated to prevent its membrane targeting. To direct the FKBP-fused G-protein regulators to membranes upon rapamycin stimulation, the FRB domain that dimerizes with FKBP was fused to a membrane targeting sequence. These constructs were co-expressed in HEK293T cells along with the BRET biosensor components. It should be noted that under these experimental conditions G-proteins or their regulators are not necessarily expressed in cells at the same levels as their native counterparts, and that G-proteins and regulators are modified by fluorescent protein (FP) tagging and truncation/mutation, respectively. Despite this limitation of the approach, the effect of different regulators on G-protein activity can be precisely interrogated and directly compared under the same experimental conditions while benchmarking against GPCR-mediated responses.
FKBP-fused constructs were expressed at comparable levels ( Figure 1-figure supplement 1). Consistent with the expectation that FKBP-fused constructs localized in the cytosol cannot effectively reach and activate G-proteins, no significant changes in BRET were observed in cells expressing the FKBP fusions in the absence of rapamycin stimulation (Figure 1-figure supplement 1). These constructs did not cause changes in the total levels of G-proteins expressed either. The exception was a modest increase in Gai-GTP BRET upon expression of Ric-8A, which was paralleled by a modest increase in the protein levels of Gai-YFP . The most likely explanation for the increased BRET is not the direct activation of G-proteins by Ric-8A, but an increase in non-specific donor-acceptor collisions due to the modest increase in BRET acceptor expression.

Non-receptor GEFs display different abilities to promote Ga-GTP and/ or free Gbg formation
Rapamycin stimulation led to rapid and robust formation of free Gbg in cells expressing either GIV or AGS1, whereas cells expressing Ric-8A did not display any response (Figure 2, top). The amplitude of the responses by the two non-receptor GEFs was comparable to that observed upon stimulation of the M4 muscarinic receptor (M4R), a prototypical Gi-activating GPCR, with an agonist concentration that elicits maximal activation in this assay format (Garcia-Marcos et al., 2020). In contrast, formation of Gai-GTP upon rapamycin stimulation was only detected in cells expressing AGS1 but not in cells expressing GIV or Ric-8A (Figure 2, top). The Gai-GTP response with AGS1 was comparable to that observed upon M4R stimulation. Consistent with a previous report , R12 GL, which has GDI activity in vitro, also led to an increase in free Gbg but caused no change in Gai-GTP levels ( Figure 2). The increase in free Gbg was somewhat smaller than that observed with GIV or AGS1. To further characterize and compare the mechanism of G-protein activation by non-receptor GEFs and GPCRs, we investigated the effect of pertussis toxin (PTX) on the Gbg responses observed upon stimulation of GIV, AGS1, or M4R. PTX ADP-ribosylates a cysteine residue in the C-terminal tail of Gai proteins, which precludes their binding to and activation by GPCRs. As expected, PTX completely suppressed the Gbg response upon M4R stimulation (Figure 2-figure supplement 1). As for the two non-receptor GEFs, PTX did not affect the Gbg response elicited by GIV, but greatly diminished the response by AGS1 (Figure 2-figure supplement 1). The lack of effect of PTX on the GIV-mediated response is consistent with the lack of involvement of the C-terminus of Gai in binding to GIV (de Opakua et al., 2017).
Together, these results highlight marked differences in G-protein activation mechanisms among non-receptor GEFs. AGS1 mimics GPCRs in that it activates proportionately Ga-GTP and free Gbg formation, and that its action is suppressed by PTX. In contrast, GIV and Ric-8A fail to promote detectable Ga-GTP formation, despite possessing GEF activity in vitro. For Ric-8A, this could be explained by previous observations that it cannot promote nucleotide exchange on G-protein heterotrimers (Tall et al., 2003), which might be the predominant G-protein species in cells (Krumins and Gilman, 2006). The lack of Gbg formation by Ric-8A would also be in agreement with this interpretation. The lack of measurable Gai-GTP formation by GIV is more puzzling for several reasons. First, the release of Gbg under the same experimental conditions demonstrates that GIV can rapidly act on Gai within G-protein heterotrimers in cells. Second, previous work has shown that GIV can promote the formation of Gai-GTP in cells by using antibodies that specifically recognize this species (Lin et al., 2014;Lopez-Sanchez et al., 2014;Midde et al., 2015). And third, even with a very similar experimental paradigm of chemically induced membrane recruitment, GIV has been shown to inhibit cAMP (Maziarz et al., 2018), presumably via inhibition of adenylyl cyclase by Gai-GTP. The next sections focus on addressing this puzzle on the mechanism of G-protein activation by GIV. Figure 2. Non-receptor guanine nucleotide exchange factors (GEFs) display different abilities to promote Ga-GTP and/or free Gbg formation. HEK293T cells expressing the components of the bioluminescence resonance energy transfer (BRET) biosensor for free Gbg (top) or Gai-GTP (bottom), the membrane-anchored FRB construct, and the indicated FKBP-fused G-protein regulators GIV GBA, AGS1*, Ric-8A*, or R12 GL were stimulated with rapamycin (0.5 mM) at the indicated time during kinetic BRET measurements. Stimulation of ectopically expressed M4 muscarinic receptor (M4R) with carbachol (100 mM) was done as a reference condition, and rapamycin stimulation of cells not expressing FKBP-fused constructs was done as a negative control. Bar graphs on the right summarize the BRET changes 90 s after addition of rapamycin or carbachol. Mean ± SD, n = 3-4. In the kinetic traces, the SD is displayed as bars of lighter color tone than data points and only in the positive direction for clarity. The online version of this article includes the following source data and figure supplement(s) for figure 2: Source data 1. Numerical data used for the upper panel (free Gbg biosensor). Source data 2. Numerical data used for the lower panel (Gai-GTP biosensor).     Garcia-Marcos. eLife 2021;10:e65620. DOI: https://doi.org/10.7554/eLife.65620 GIV-CT has the same G-protein activating properties as GIV GBA motif One potential caveat of the experiments with GIV above is that the construct used contained only its GBA motif. Although it is unlikely that this would explain the lack of Gai-GTP formation, because the GBA motif is sufficient to promote nucleotide exchange in vitro (de Opakua et al., 2017), this issue was investigated more thoroughly by using a larger GIV construct. Experiments in this assay format with full-length GIV are not feasible because the protein is >250 KDa and contains multiple domains that associate with membranes or cytoskeletal components. Instead, a C-terminal region of 210 amino acids (1660-1870, GIV-CT) was used. GIV-CT not only fully recapitulates the properties of GIV GBA motif in activating G-proteins in vitro (de Opakua et al., 2017), but also recapitulates the properties of full-length GIV in promoting G-protein-dependent signaling in cells (Ma et al., 2015a;Midde et al., 2015). Rapamycin-induced recruitment of GIV-CT induced an increase in free Gbg levels similar to that caused by GIV GBA (Figure 2-figure supplement 2). Also like GIV GBA, it failed to elicit any detectable increase in Gai-GTP, suggesting that the lack of Gai-GTP response is not an inherent defect of the shorter construct.
cAMP dampening by GIV's GBA motif is blocked upon Gbg scavenging Suppression of cAMP through direct inhibition of adenylyl cyclase activity is what originally defined the Gi subfamily of G-proteins. Although this is widely attributed to the action of Gai-GTP on adenylyl cyclases, Gbg can also directly modulate the production of cAMP by adenylyl cyclases (Sadana and Dessauer, 2009;Sunahara et al., 1996). GIV has been previously shown to decrease cAMP levels in cells (Maziarz et al., 2018;Midde et al., 2015), and the results presented above ( Figure 2) indicate that it promotes the formation of detectable levels of Gbg but not of Gai-GTP. Together, the above prompted the investigation of whether GIV-induced cAMP dampening is mediated though Gbg. For this, the effect of GIV recruitment to membranes on forskolin-induced cAMP was determined with or without co-expression of the C-terminal region of GRK2 (GRK2ct) (Figure 3). GRK2ct binds with high affinity to free Gbg subunits and precludes their binding to effectors (Koch et al., 1994). Consistent with previous findings by Maziarz et al., 2018, GIV recruitment to membranes led to a decrease in cAMP ( Figure 3). This effect of GIV was efficiently suppressed by expression of GRK2ct (Figure 3), suggesting that it is primarily mediated by the formation of free Gbg subunits rather than by Gai-GTP.
Recruitment of GIV's GBA motif to RTKs promotes Gbg release but not Gai-GTP formation Kalogriopoulos et al., 2020 have recently proposed that GIV facilitates EGFR-mediated phosphorylation of Gai by binding simultaneously to the inactive G-protein and the active, auto-phosphorylated RTK. Formation of this complex leads to Gai phosphorylation by EGFR, which in turn promotes GTP loading by accelerating nucleotide exchange ( Figure 3-figure supplement 1, top). This posits that a mechanism by which GIV promotes Gai-GTP formation in cells is through GBA motif-dependent recruitment of Gai to the vicinity of EGFR. To directly test this model (Figure 3-figure supplement 1, top), Gai-GTP BRET was determined upon EGF stimulation in HEK293T cells expressing EGFR and GIV. In initial experiments with cells expressing GIV-CT, which contains both the G-protein and the RTK binding regions (Lin et al., 2014), no Gai-GTP or free Gbg formation was detected upon EGF stimulation (data not shown), suggesting that recruitment of GIV to EGFR is inefficient under these conditions. As an alternative to overcome this limitation, we addressed the impact of GIV-dependent recruitment of Gai to EGFR by fusing GIV's GBA motif to the adaptor protein Grb2, which is efficiently recruited to active EGFR (Lowenstein et al., 1992; Figure 3-figure supplement 1). In cells expressing the Grb2-GBA fusion, EGF stimulation led to an increase of free Gbg but not of Gai-GTP (Figure 3-figure supplement 1). The Gbg response was not recapitulated when a Grb2-GBA construct bearing the Gai binding-deficient mutation F1685A (FA) of GIV (de Opakua et al., 2017;Garcia-Marcos et al., 2009) was used (Figure 3-figure supplement 1), indicating that the observed response is specifically caused by GIV's GBA motif and not by other EGFR triggered signaling events. These results show that, while an active EGFR-GIV complex engages Gi proteins to promote the release of Gbg, it is still inefficient in promoting the formation of Gai-GTP.  Gbg release by GIV is insensitive to cellular GTP depletion It is possible that (i) GIV promotes the formation of Gai-GTP below the detection levels of the BRET assay used above, and that (ii) this in turn contributes to the formation of larger levels of free Gbg. The former is suggested by previous evidence of GIV-dependent Gai-GTP formation in cells by using an antibody-based approach (Maziarz et al., 2018;Midde et al., 2015), which might be more sensitive than the BRET assay above. The latter is suggested by the observation that R12 GL promotes a smaller increase of free Gbg than GIV (Figure 2), even though R12 GL has an affinity for Gai about 10 times higher than that of GIV (de Opakua et al., 2017). It is therefore conceivable that GIV utilizes mechanisms other than just physical displacement by mass action to achieve formation of free Gbg more efficiently than R12 GL. For example, after displacing Gbg from Gai, GIV might weakly promote Gai-GTP formation to sustain the dissociated status of Gbg, which R12 GL could not because it is a GDI. This point was addressed by investigating the requirement of GTP for GIV-mediated Gbg responses using a nucleotide depletion protocol in semi-permeabilized cells previously described by Qin et al., 2011;Qin et al., 2008. BRET responses triggered by different regulators were compared in cells depleted of nucleotides or replenished with near-physiological levels of GTP (0.25 mM). GIV-induced Gbg responses were very similar in the presence or absence of added GTP after the nucleotide depletion protocol (Figure 4). Similar observations were made for R12 GLinduced Gbg BRET responses, which are not expected to require GTP. In contrast, agonist-stimulated GPCR Gbg BRET responses were larger in the GTP-replenished condition (Figure 4). Although it is unclear if the GPCR response observed in the absence of GTP addition is due to incomplete nucleotide depletion and/or a G-protein rearrangement that occurs upon engagement with active GPCRs in the absence of nucleotides Rasmussen et al., 2011), these results indicate that Gbg release induced by GIV is largely independent of the presence of physiological levels of GTP.
GIV GBA does not suppress GPCR-mediated Ga-GTP formation like R12 GL Data presented above shows that GIV GBA and R12 GL display similar G-protein signaling properties in cells despite having striking different behavior in vitro-that is, the former is a GEF and the latter is a GDI. Next, the effects of GIV GBA and R12 GL on Ga-GTP formation were compared under conditions of GPCR stimulation, which represent a regime of high nucleotide exchange not present in the previously investigated resting conditions. After stimulation of M4R with carbachol, addition of rapamycin led to a rapid drop of Gai-GTP in cells expressing R12 GL compared to controls, whereas no significant change in Gai-GTP was observed in cells expressing GIV GBA ( Figure 5). These results suggest that the GDI activity of R12 GL efficiently sequesters monomeric Ga-GDP, reducing the pool of Ga(GDP)-Gbg trimeric substrate used by the GPCR to sustain the higher steady-state levels of Ga-GTP achieved upon stimulation. In contrast, GIV GBA, which is even more efficient than R12 GL in dissociating Ga-Gbg complexes (Figure 2), does not seem to significantly prevent the utilization of Ga(GDP)-Gbg by GPCRs, suggesting that it does not lead to sequestration of monomeric Ga-GDP. As opposed to what occurs with a GDI like R12 GL, which locks Ga in its GDP-bound state, GIV-bound Ga can exchange nucleotides. Then, GIV would dissociate from the G-protein upon transient formation of Gai-GTP, which would be rapidly converted into Ga-GDP to replenish Ga(GDP)-Gbg levels.

GIV GBA does not hinder GPCR-mediated activation of G-proteins
The sequestration model proposed above was further investigated by characterizing how different cytoplasmic regulators affect GPCR-mediated G-protein activation. Essentially, three qualitative Source data 1. Numerical data used for the lower panel (Gai-GTP biosensor). Figure supplement 1. Recruitment of GIVGBA motif to RTKs promotes Gbg release but not Gai-GTP formation. Figure supplement 1-source data 1. Numerical data used for the upper graphs (free Gbg biosensor).

Figure supplement 1-source data 2. Numerical data used for the lower graphs (Gai-GTP biosensor).
Garcia-Marcos. eLife 2021;10:e65620. DOI: https://doi.org/10.7554/eLife.65620 . Gbg release by GIV is insensitive to cellular GTP depletion. HEK293T cells expressing the components of the BRET biosensor for free Gbg, the membrane-anchored FRB construct, and FKBP-fused GIV GBA or R12 GL were depleted of nucleotides (black) or replenished with GTP (0.25 mM, red) as indicated in 'Materials and methods'. Cells were stimulated with rapamycin (0.5 mM) at the indicated time during kinetic BRET measurements. Stimulation of ectopically expressed M4R with carbachol (100 mM) was done as a reference condition, and rapamycin stimulation of cells not expressing FKBP-fused constructs was done as a negative control. The bar graph on the bottom summarizes the BRET changes 90 s after addition of rapamycin or carbachol. Mean ± SD, n = 3-4. In the kinetic traces, the SD is displayed as bars of lighter color tone than data points and only in the positive direction for clarity. The online version of this article includes the following source data for figure 4: Source data 1. Numerical data used for the graphs. scenarios were proposed and tested considering that the availability of Ga(GDP)-Gbg complexes is limiting their utilization by GPCRs to generate Gai-GTP and free Gbg ( Figure 6A). In each scenario, the action of cytoplasmic regulators is triggered before GPCR stimulation, which leads to a new dynamic equilibrium between Ga(GDP)-Gbg and active Ga/Gbg species prior to receptormediated activation. In scenario one, the GDI R12 GL reduces the availability of Ga(GDP)-Gbg by dissociating Ga from Gbg and locking it in the Ga-GDP state. In two, GIV GBA does not reduce the availability of Ga(GDP)-Gbg because the association with Ga-GDP can be reversed upon nucleotide exchange. In three, the GEF AGS1 reduces the availability of Ga(GDP)-Gbg by competing for it with GPCRs to independently generate Ga-GTP and free Gbg through multiple turnover cycles. Consistent with these proposed models, the formation of Gai-GTP and free Gbg upon M4R stimulation with carbachol were reduced by pretreatment with rapamycin in cells expressing R12 GL or AGS1 but not in cells expressing GIV ( Figure 6B,C). The total Gbg BRET change observed after GIV recruitment and GPCR stimulation (time 240 s, Figure 6C) was larger than in any of the other conditions, indicating that the reduced GPCR response by R12 GL or AGS1 was not due to reaching a maximum signal. Furthermore, if the association of GIV GBA with Ga-GDP can be reverted as proposed, one would also expect that its ability to generate free Gbg would be hampered after GPCR stimulation due to competitive removal of available Ga (GDP)-Gbg. In other words, once a GPCR is activated, formation of Ga(GDP)-GIV (and subsequent generation of free Gbg) would be disfavored because the equilibrium of G-protein complexes is shifted toward formation of receptor-Ga(GDP)-Gbg. Indeed, the Gbg response to GIV GBA was diminished in cells in which M4R had been pre-stimulated with carbachol ( Figure 6figure supplement 1). Taken together, these results suggest that GIV, despite efficiently triggering G-protein signaling via Gbg, does oppose activation of G-proteins by GPCRs.

Discussion
This work introduces an experimental paradigm to investigate the regulation of G-proteins in cells with high precision and for a wide range of regulators beyond GPCRs. The experimental framework presented here is poised to complement other approaches that have traditionally been used to study G-protein signaling, like reductionist biochemical assays with purified proteins and genetic manipulations for more complex systems like cultured cells or whole organisms. Experiments with purified proteins allow gaining detailed insights into molecular mechanisms but can suffer from lack of physiological context, whereas genetic manipulations can reveal functionality in more physiological contexts but can fall short in delivering mechanistic detail at the molecular level. The approach presented here leverages some advantages of reductionism, in that it permits studying isolated G-protein biochemical events like nucleotide binding status and subunit dissociation upon Figure 5. R12 GL, but not GIV GBA, suppresses GPCRmediated Ga-GTP formation. HEK293T cells expressing the components of the BRET biosensor for Gai-GTP, the membrane-anchored FRB construct, M4R, and FKBP-fused GIV GBA or R12 GL were sequentially stimulated with carbachol (100 mM) and rapamycin (0.5 mM) at the indicated times during kinetic BRET measurements. Mean ± SD, n = 4. SD is displayed as bars of lighter color tone than data points and only in the positive direction for clarity. The online version of this article includes the following source data for figure 5: Source data 1. Numerical data used for the graphs.
Garcia-Marcos. eLife 2021;10:e65620. DOI: https://doi.org/10.7554/eLife.65620 modulation by specific protein regulators, but within the more physiological context of the cell. In this sense, it could be considered a 'cell-based reductionist' approach. The benefits of this are demonstrated here by the discovery that G-protein signaling modes in cells can differ greatly among Figure 6. R12 GL and AGS1*, but not GIV GBA, hinder the activation of G-proteins by a GPCR. (A) Proposed models to explain the mode of action of different cytoplasmic regulators of G-proteins on GPCR-mediated activation. (B, C) HEK293T cells expressing the components of the BRET biosensor for Gai-GTP (B) or free Gbg (C), M4R, the membrane-anchored FRB construct, and the indicated FKBP-fused G-protein regulators (R12 GL, GIV GBA, or AGS1*) were sequentially stimulated with rapamycin (0.5 mM) and carbachol at the indicated times during kinetic BRET measurements. Stimulation of cells not expressing FKBP-fused constructs with rapamycin was done as a control. Bar graphs on the top summarize the BRET changes 90 s after addition of carbachol. Mean ± SD, n = 4. In the kinetic traces, the SD is displayed as bars of lighter color tone than data points and only in the positive direction for clarity. The online version of this article includes the following source data and figure supplement(s) for figure 6: Source data 1. Numerical data used for the graphs on the left (Gai-GTP biosensor). Source data 2. Numerical data used for the graphs on the right (free Gbg biosensor).  proteins with similar G-protein regulatory functions in vitro, including differences between receptor and non-receptor GEFs or even among different non-receptor GEFs (see further discussion below). Also, the interplay between GPCRs and different cytoplasmic regulators in controlling G-protein activity in cells was further clarified. Like with biochemical reductionist approaches, a limitation of the studies presented here and the approach in general if considered alone is that they still rely on non-native conditions, like the use of overexpression, protein fragment fusions, or artificial means to alter the subcellular localization of proteins. It is in the context of complementarity with other wellestablished approaches like in vitro biochemistry and genetics that this limitation is outweighed by the additional mechanistic insights it can provide.
The in-depth characterization of the mechanisms by which non-GPCR proteins of the GBA family trigger G-protein signaling in cells exemplifies well the additional insights that can be gained through the approach presented in this work. On one hand, it was known from in vitro biochemical experiments that GBA motifs have GEF activity on Gai proteins and that they also promote the dissociation of Gbg from Gai-GDP Coleman et al., 2016;de Opakua et al., 2017;Garcia-Marcos et al., 2011a;Garcia-Marcos et al., 2010;Garcia-Marcos et al., 2009;Garcia-Marcos et al., 2012;Garcia-Marcos et al., 2011c;Marivin et al., 2019;Maziarz et al., 2018). On the other hand, genetic approaches had also established that the GBA motif of some proteins, like GIV and DAPLE, controls G-protein signaling in cells or whole organisms Garcia-Marcos et al., 2011a;Garcia-Marcos et al., 2010;Garcia-Marcos et al., 2009;Garcia-Marcos et al., 2012;Landin Malt et al., 2020;Leyme et al., 2016;Leyme et al., 2017;Leyme et al., 2015;Lin et al., 2014;Lo et al., 2015;Lopez-Sanchez et al., 2014;Ma et al., 2015b;Marivin et al., 2019;Midde et al., 2015;Sasaki et al., 2015). Thus, an unresolved question so far had been the relative contribution of the two potentially overlapping mechanism of G-protein activation-mediated GBA motifs, that is, generation of Gai-GTP and formation of free Gbg. A key conclusion of the studies presented here (Figure 2, Figure 4) is that GIV, and most likely other GBA proteins, activates G-protein signaling in cells primarily through the formation of free Gbg rather than through the formation of Gai-GTP, despite its GEF activity in vitro. GIV enhances nucleotide exchange in vitro approximately 2.5-to 3-fold (Garcia-Marcos et al., 2011a;Garcia-Marcos et al., 2010;Garcia-Marcos et al., 2009;Garcia-Marcos et al., 2012), which is very similar to the approximately 3-fold enhancement mediated by AGS1 under similar conditions in vitro (Cismowski et al., 2000). In contrast, the studies described here reveal that AGS1 efficiently triggers the formation of Gai-GTP in cells, whereas GIV does not (Figure 2). Similarly, the enhancement of nucleotide exchange on Gi by a GPCR (e.g. a2 adrenergic receptor) in vitro is 3-to 6-fold (Cerione et al., 1986;Kurose et al., 1991), which is stronger than that of GIV or AGS1 but still within the same order of magnitude. Thus, the results presented here indicate that the relative GEF activity in vitro across different G-protein activators does not correlate well with their ability to generate Ga-GTP in cells. These observations not only provide mechanistic insights into G-protein activation by GBA proteins that were not evident from experiments using other approaches, but also highlight the importance of elucidating the molecular mechanisms of G-protein regulation in a cellular context.
The conclusion that GIV promotes G-protein signaling in cells primarily through the formation of free Gbg prompts the re-evaluation of previous signaling studies in cells. In fact, the vast majority of signaling readouts regulated by the GBA motif of GIV (or DAPLE) have been shown or are known to be controlled by Gbg rather than by Gai-GTP. This includes activation of the PI3K-Akt axis Bhandari et al., 2015;Garcia-Marcos et al., 2010;Garcia-Marcos et al., 2009;Garcia-Marcos et al., 2012;Gupta et al., 2016;Leyme et al., 2016;Leyme et al., 2017;Leyme et al., 2015;Ma et al., 2015a), p114RhoGEF-mediated activation of RhoA (Marivin et al., 2019), or Rac1 activation . Although GBA-mediated inhibition of adenylyl cyclase Maziarz et al., 2018;Midde et al., 2015) could have been explained as an effect of Gai-GTP, evidence presented here (Figure 3) strongly suggests that this is indeed mediated by Gbg as well. Thus, although previous evidence has identified detectable levels of Gai-GTP in cells upon GIV action (Lin et al., 2014;Lopez-Sanchez et al., 2014;Midde et al., 2015), these are probably much lower than those achieved upon activation of a GPCR and insufficient to drive robust signaling directly. Thus, the functional role of the weak GEF activity of proteins with a GBA motif remains to be elucidated, whereas efficient release of Gbg subunits seems to be main mechanism by which this type of G-protein activator promotes signaling in cells.
Results shown here ( Figure 5, Figure 6) also indicate that proteins with a GBA motif differ mechanistically from other regulators that promote the dissociation of Gbg from G-protein heterotrimers, like GDIs that contain a GoLoco motif. This in turn might be important for the interplay between GBA proteins and GPCRs in regulating G-proteins. The data suggest that activation of G-proteins by GIV, which in many cases is triggered by surface receptors different from GPCRs Leyme et al., 2015;Lopez-Sanchez et al., 2014), might operate without opposing efficient GPCR-mediated G-protein activation. In contrast, GDIs with a GoLoco motif efficiently suppress GPCR-mediated activation of G-proteins. Interestingly, the artificial system implemented here to recruit GIV to membranes mimics the recruitment of native GIV from the cytosol to membranes when it binds non-GPCR surface receptors upon ligand stimulation Leyme et al., 2015), which might be the mechanism by which GIV action on G-proteins is controlled under native conditions . Nevertheless, the role of GIV in the interplay between GPCR and non-GPCR receptors in G-protein regulation needs to be characterized in more detail in the future.
Beyond insights gained in the understanding of GBA-mediated mechanisms of G-protein signaling regulation, the present study also provides other useful information. For example, it was shown that pertussis toxin is not a generic inhibitor of all mechanisms of Gi activation or a specific inhibitor of GPCR-mediated activation of Gi proteins (Figure 2-figure supplement 1), which warranties caution in the interpretation of past and future experiments with this widely used reagent. In addition, the chemogenetic tools presented here could be easily adapted for other applications, like synthetic biology approaches to turn different modes of G-protein signaling ON and OFF.
In summary, the combination of chemogenetics and optical biosensors presented here has demonstrated the potential to become an experimental paradigm to expand how we study and understand signal transduction mechanisms mediated by heterotrimeric G-proteins. Continued on next page University, Philadelphia, PA) (Irannejad and Wedegaertner, 2010) into SdaI/SmaI sites of a pCS2 + plasmid that places a 6xMyc tag sequence at the N-terminus of the insert (Marivin et al., 2019).

G-protein live-cell BRET measurements
HEK293T cells (ATCC, cat# CRL-3216) were grown at 37˚C, 5%CO 2 in high-glucose Dulbecco's modified eagle medium (DMEM) supplemented with 10% FBS, 100 U/ml penicillin, 100 mg/ml streptomycin, and 1% L-glutamine. HEK293T cells were not authenticated by STR profiling or tested for mycoplasma contamination. Approximately 400,000 cells/well were seeded on 6-well plates coated with 0.1% gelatin and transfected~24 hr later using the calcium phosphate method. For experiments aimed at detecting free Gbg, cells were transfected with the following amounts of plasmid DNA per well: 1 mg for Gai3, 0.2 mg for VC-Gb 1 , 0.2 mg VN-Gg 2 , and 0.1 mg of mas-GRK3ct-Nluc. For experiments aimed at detecting Gai-GTP, cells were transfected with the following amounts of plasmid DNA per well: 1 mg for Gai3-YFP, 0.2 mg for Gb 1 , 0.2 mg Gg 2 , and 0.1 mg of mas-KB-1753-Nluc. For either Gbg or Gai-GTP measurements, cells were co-transfected with the following amounts of plasmid DNA per well: 3 mg for Lyn11-FRB, 0.2 mg for M4R, 0.5 mg for FKBP-GIV GBA, 0.125 mg for FKBP-AGS1*, 0.05 mg for FKBP-Ric8A*, 0.1 mg for FKBP-R12 GL, and 0.5 mg for FKBP-GIV-CT. Total DNA amount per well was equalized by supplementing with empty pcDNA3.1 as needed. For experiments shown in Figure 3-figure supplement 1, Lyn11-FRB and FKBP-fused constructs were omitted and the following amounts of plasmid DNA transfected instead: 1 mg for EGFR and 2 mg for Grb2-GBA. Approximately 18-24 hr after transfection, cells were washed with PBS, harvested by gentle scraping, and centrifuged for 5 min at 550 Â g. Cells were resuspended in assay buffer (140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 0.37 mM NaH 2 PO 4 , 20 mM HEPES pH 7.4, 0.1% glucose) at a concentration of approximately 1 million cells/ml; 25,000-50,000 cells were added to a white opaque 96-well plate (Opti-Plate, Perkin Elmer) and mixed with the nanoluciferase substrate Nano-Glo (Promega, cat# N1120, final dilution 1:200) for 2 min before measuring luminescence signals in a POLARstar OMEGA plate reader (BMG Labtech) at 28˚C. Luminescence was measured at 460 ± 40 and 535 ± 10 nm, and BRET was calculated as the ratio between the emission intensity at 535 ± 10 nm divided by the emission intensity at 460 ± 40 nm. For kinetic BRET measurements, luminescence signals were measured every 0.24 s for the duration of the experiment. Reagents were added to the wells during live measurements using injectors. Kinetic measurement data are presented as the BRET change relative to the baseline signal (the average BRET ratio of the 30 s prestimulation). For endpoint measurements shown in Figure 1-figure supplement 1, data is presented as raw BRET ratios (535 nm luminescence/460 nm luminescence) of unstimulated cells. For Figure 2-figure supplement 1, PTX treatments consisted of overnight incubations with 0.2 mg/ml of the toxin. For Figure 4, cells were nucleotide-depleted by following a previously described protocol Qin et al., 2008). Procedures were as described above except that cells were resuspended in a different assay buffer (140 mM potassium gluconate, 5 mM KCl, 10 mM HEPES, 1 mM EGTA, 0.3 mM CaCl 2 , 1 mM MgCl 2 , pH 7.2), and treated with 1000 U/ml of a-hemolysin and 5 mM KCN for 10 min before the start of the measurements to semi-permeabilize cells and block nucleotide synthesis, respectively. As indicated in the figures, 0.25 mM GTP was added in some cases for 2 min before stimulation with rapamycin or carbachol.
At the end of some BRET experiments, a separate aliquot of the same pool of cells used for the luminescence measurements was centrifuged for 1 min at 14,000 Â g and pellets stored at À20˚C for subsequent immunoblot analysis (see 'Protein electrophoresis and Immunoblotting' section below). cAMP live-cell BRET measurements HEK293T cells were seeded and transfected using the calcium phosphate method as in 'G-protein live-cell BRET measurements' section but using the following amounts of plasmid DNA per well: 3 mg for Lyn11-FRB, 0.5 mg for FKBP-GIV GBA, 0.05 mg for Nluc-EPAC-VV, and 2 mg for GRK2ct. Total DNA amount per well was equalized by supplementing with empty pcDNA3.1 as needed. Luminescence measurements were also carried out as described in 'G-protein live-cell BRET measurements' section, except that signals were measured every 4 s instead of every 0.24 s. Results were presented as the inverse of the BRET ratio after subtraction of the basal BRET signal measured for 60 s before any stimulation (BRET change À1 ). Inhibition of forskolin-induced cAMP after 15 min was determined by calculating the difference in BRET changes with and without rapamycin addition after subtraction of a baseline signal measured in parallel with unstimulated cells. Samples for immunoblotting were prepared as in 'G-protein live-cell BRET measurements' and processed as described in 'Protein electrophoresis and Immunoblotting' section below.