Receptor-regulated Interaction of Activator of G-protein Signaling-4 and Gαi*

Activator of G-protein signaling-4 (AGS4), via its three G-protein regulatory motifs, is well positioned to modulate G-protein signal processing by virtue of its ability to bind Gαi-GDP subunits free of Gβγ. Apart from initial observations on the biochemical activity of the G-protein regulatory motifs of AGS4, very little is known about the nature of the AGS4-G-protein interaction, how this interaction is regulated, or where the interaction takes place. As an initial approach to these questions, we evaluated the interaction of AGS4 with Gαi1 in living cells using bioluminescence resonance energy transfer (BRET). AGS4 and Gαi1 reciprocally tagged with either Renilla luciferase (RLuc) or yellow fluorescent protein (YFP) demonstrated saturable, specific BRET signals. BRET signals observed between AGS4-RLuc and Gαi1-YFP were reduced by G-protein-coupled receptor activation, and this agonist-induced reduction in BRET was blocked by pertussis toxin. In addition, specific BRET signals were observed for AGS4-RLuc and α2-adrenergic receptor-Venus, which were Gαi-dependent and reduced by agonist, indicating that AGS4-Gαi complexes are receptor-proximal. These data suggest that AGS4-Gαi complexes directly couple to a G-protein-coupled receptor and may serve as substrates for agonist-induced G-protein activation.

Activators of G-protein signaling (AGS) 3 proteins were identified using a yeast-based functional screen of mammalian cDNA libraries for cDNAs that activated G-protein signaling in the absence of a GPCR (1)(2)(3)(4). Group II AGS proteins all contain at least one G-protein regulatory (GPR) motif (3,5) (also termed the GoLoco motif (6)), a 20 -25-amino acid motif that binds and stabilizes the GDP-bound conformation of G␣ i / o / t and competes with G␤␥ for G␣ binding (reviewed in Ref. 5). Proteins with multiple GPR motifs can bind to multiple G␣ subunits simultaneously, which presents a unique opportunity to act as a scaffold to organize a signaling complex (7,8).
Functional studies indicate crucial roles for GPR proteins beginning with the original observations in model organisms describing a role for GPR proteins and their interaction with G-proteins in asymmetric cell division (5,9). Additional functional studies with GPR proteins indicate further functional diversity with roles observed in blood pressure control, fat deposition and energy expenditure, neuronal outgrowth, drug addiction and relapse behavior, autophagy, G-protein-coupled inwardly rectifying potassium channel regulation, and transport of membrane proteins to the cell surface (10 -18). These observations indicate crucial functionality of the GPR motif in biological systems and implicate G␣ i -GPR complexes in the regulation of G-protein signaling in unexpected, albeit poorly understood ways. In the context of the group II AGS proteins, which contain multiple GPR motifs, many outstanding questions remain to be addressed. Chief among them is what regulates the formation and disassembly of GPR-G␣ i complexes? Is their interaction with G-protein influenced by GPCR activation or other signals? AGS4 was identified in the yeast based functional screen for receptor-independent G-protein activators from a human prostate leiomyosarcoma cDNA library (1), and apart from initial reports describing its interaction with G␣ i subunits from cell lysates and purified proteins in vitro, little is known regarding its interaction with G␣ i in the intact cell or how this interaction is regulated. Although AGS3 and AGS5/Leu-Gly-Asn repeat-enriched protein (LGN) both contain seven tetratricopeptide repeats upstream of their four GPR motifs and are broadly expressed in many tissues, AGS4 has a much different domain organization with a proline-rich N terminus followed by three GPR motifs that are somewhat dissimilar as compared with AGS3 and AGS5/LGN (1,19). AGS4 also appears to be more restricted in its expression to the immune system. 4 In this study, we report the interaction of AGS4 with G␣ i in the intact cell as determined by bioluminescence resonance energy transfer (BRET) and its regulation by cell surface GPCRs. In addition, the data indicate that AGS4-G␣ i complexes are receptor-proximal as specific BRET signals were observed between AGS4 and the ␣ 2A -adrenergic receptor (␣ 2A -AR), which were G␣ i -dependent and reduced by an ␣ 2A -AR agonist. These data suggest that AGS4-G␣ i complexes directly couple to a GPCR and may serve as substrates for receptor-induced G-protein activation.
Cell Culture and Transfection-HEK-293 cells were maintained in Dulbecco's minimal essential medium (high glucose, without phenol red) containing 5% fetal bovine serum, 2 mM glutamine, 100 units/ml penicillin, and 100 mg/ml streptomycin. Cells were grown in the presence of 5% CO 2 at 37°C in a humidified incubator. For transfection, 8 ϫ 10 5 cells/well were seeded on 6-well plates and cultured overnight at 37°C. BRET donor and acceptor plasmids were used for transfection with PEI (1 mg/ml in distilled H 2 O) at a DNA:PEI ratio of 1:4. PEI and plasmid DNA were diluted in separate tubes with 100 l of serum-free medium. DNA and PEI solutions were vortexed at maximum speed for 3-5 s and incubated for 15 min at room temperature prior to addition to the cells. Cells were incubated for 48 h prior to collection for experiments. Cell lysates and immunoblotting were performed as described previously (24).
BRET-Initial experiments were performed to optimize the BRET system for AGS4-G␣ i1 interactions and to ensure the specificity of observed signals. All studies involved saturation BRET analysis, altering donor/acceptor ratios and/or time course analysis. Forty-eight hours after transfection, the cells were washed once with phosphate-buffered saline and harvested with Tyrode's solution (140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 0.37 mM NaH 2 PO 4 , 24 mM NaHCO 3 , 10 mM HEPES, pH 7.4, and 0.1% glucose (w/v)). Cells were distributed in triplicate at 1 ϫ 10 5 cells/well into gray 96-well plates. Fluorescence and luminescence signals were measured with a TriStar LB 941 plate reader (Berthold Technologies, Oak Ridge, TN). Total fluorescence (excitation, 485 nm; emission, 535 nm) was first measured to quantify acceptor expression. The luciferase substrate coelenterazine H (5 M final concentration) was then added, and luminescence was measured (donor, 480 Ϯ 20 nm; acceptor, 530 Ϯ 20 nm). Net BRET values were determined by first calculating the 530 Ϯ 20:480 Ϯ 20 nm ratio and then subtracting the background BRET signal determined from cells transfected with the Renilla luciferase (RLuc) expression vector phRLuc N3 alone. Spectral measurements were conducted with the protocol described above using a SpectraMax M5 plate reader (Molecular Devices, Sunnyvale, CA). BRET saturation curves and statistical analyses were measured using GraphPad Prism (GraphPad Software, San Diego, CA). Data were analyzed by analysis of variance with significant differences between groups determined by Tukey's post-hoc test.

RESULTS AND DISCUSSION
Key questions in the field and for AGS4 in particular are what regulates the formation and disassembly of AGS4-G␣ i complexes and is the AGS4-G-protein interaction influenced by GPCR activation or other signals? As an initial approach to address these questions, we used BRET with contingent binding proteins tagged with RLuc or yellow fluorescent protein (YFP). The enzymatic oxidation by luciferase of substrates such as coelenterazine and subsequent non-radiative emission can excite YFP if the two proteins are in close proximity (Ͻ 100 Å). The strength of the BRET signal for two proteins fused with RLuc and YFP, respectively, depends upon distance between the fluorophores, their relative orientations, and the relative expression levels of donor (RLuc) and acceptor (YFP). Specific interactions exhibit saturation of BRET signals using a constant amount of the luciferase donor and increasing amounts of the YFP acceptor (25). AGS4 was tagged with either YFP or RLuc, and YFP or RLuc was inserted into the loop connecting helices ␣B and ␣C of the helical domain in G␣ i1 , which confers nucleotide binding and hydrolysis properties similar to the untagged protein (20,22).
Analysis of different AGS4 and G␣ i BRET donor/acceptor pair combinations revealed that the AGS4-RLuc-G␣ i1 -YFP (YFP at position 122) BRET pair yielded the strongest BRET signal and the highest signal-to-noise ratio (data not shown). The higher BRET signals detected by C-terminal tagged AGS4-RLuc and G␣ i1 -YFP likely reflect the ability of the donor AGS4-RLuc to simultaneously bind more than one acceptor G␣ i1 -YFP due to the presence of three GPR motifs in AGS4. Robust, specific, and saturable BRET signals were observed between AGS4-RLuc and G␣ i1 -YFP (Fig. 1, A and B). BRET signals were not observed in cells expressing AGS4-Q/A, which contains Gln-Ala mutations in each of the three GPR motifs in AGS4, rendering it unable to bind G␣ i (26) (Fig. 1, A and B), thus confirming the specificity of the BRET signal. AGS4-RLuc-G␣ i1 -YFP BRET signals were also decreased following co-expression of G␤␥ subunits, consistent with the ability of G␤␥ subunits to compete with GPR motifs for G␣ binding (8, 27) (Fig. 1B). AGS4-RLuc-G␣ i -YFP BRET signals were markedly reduced upon introduction of the Q204L mutation in G␣ i -YFP, which alters GTP hydrolysis consistent with the known preference of GPR motifs for GDP-bound G␣ i subunits. The N149I mutation, which renders G␣ i incapable of binding GPR motifs (28), also predictably reduced the AGS4-G␣ i BRET signal (Fig.  1C). Interestingly, treatment of cells with pertussis toxin, which ADP-ribosylates a cysteine residue in G␣ i / o subunits four amino acids from the C terminus and renders G␣ incapable of being activated by GPCRs, had no effect AGS4-RLuc-G␣ i -YFP BRET signals (Fig. 1C).
Truncations of AGS4 revealed that the AGS4-RLuc-G␣ i1 -YFP BRET signals require the AGS4 C-terminal GPR domain (AGS4-CT-Leu 57 -Cys 160 ) (Fig. 1D). However, although no BRET signals were observed between the AGS4 N terminus (AGS4-NT-Met 1 -Ser 56 ) and G␣ i -YFP, the reduction in the overall magnitude of the BRET signals of AGS4-CT-RLuc as compared with full-length AGS4-RLuc suggests that the N terminus of AGS4 may influence the ability of AGS4 to interact with G␣ i , consistent with the idea that residues outside of the core GPR motif may influence the GPR-G␣ i interaction (1,19,26,29).
As an initial approach to define the relative influence of each of the GPR motifs in AGS4 on G␣ i binding, we tested AGS4-RLuc constructs with the Q/A substitution in each of the AGS4-GPR motifs (Q80A, Q122A, and Q151A) in AGS4-RLuc-G␣ i -YFP BRET experiments (Fig. 1E). These data are interesting in a number of aspects. First, individually mutated AGS4-GPR motifs show a progressive decrease in G␣ i1 -YFP BRET signals with GPRIII-Q/A having the most significant effect. In addition, AGS4-RLuc constructs with mutations in GPRIII in combination with either GPRI (GPRI,III-Q/A) or GPRII (GPRII-III-Q/A) have significantly lower BRET signals with G␣ i1 -YFP than AGS4-Q/A mutations in the first two GPR motifs (GPRI-II-Q/A). Taken together these data suggest that GPR-III is important for G␣ i binding in cells. Alternatively, the decreased BRET signals generated in GPR-III Q/A mutants may reflect the proximity of the C-terminal RLuc tag to GPR-III and increased resonance energy transfer to G␣ i -YFP. Secondly, Q/A mutations in both GPR-I and GPR-III of AGS4 (GPRI,III) still yield significant G␣ i -YFP BRET signals with GPR-II as the sole GPR motif, suggesting that in the intact cell, GPR-II does indeed contribute to G␣ i binding consistent with earlier observations (1). This is in contrast to a previous report in which a glutathione S-transferase fusion protein containing AGS4-GPRII was essentially inactive as a guanine nucleotide dissociation inhibitor for G␣ i1 unless Ala 121 was changed to Asp (19). We then sought to determine the influence of GPCR activation on the AGS4-G␣ i interaction. Co-expression of the ␣ 2Aadrenergic receptor had no effect on AGS4-RLuc-G␣ i -YFP BRET signal generation ( Fig. 2A), and BRET signals were detected at levels of G␣ i1 -YFP that were similar to the endogenous level of G␣ i1 detected by immunoblotting (Fig. 2B, lower  panel). However, when the ␣ 2 -AR agonist UK14304 was added, an ϳ30% reduction in AGS4-RLuc-G␣ i -YFP BRET signal was observed (Fig. 2, A and B). This agonist-mediated effect was dose-dependent (Fig. 2C), occurred within 2 min of agonist treatment, and persisted for up to 30 min (Fig. 2D). The UK14304-mediated reduction in AGS4-RLuc-G␣ i1 -YFP BRET was blocked by the ␣ 2 -AR antagonist rauwolscine (Fig. 2E). Similar reductions in AGS4-RLuc-G␣ i1 -YFP BRET were observed upon co-expression of CXCR4 and the mu-opioid receptor (MOR) after treatment with CXCL12 and [D-Ala 2 , N-MePhe 4 , Gly-ol]-enkephalin, respectively (data not shown). Similar reductions in BRET signals were observed between the GPR protein AGS3 fused to Renilla luciferase and G␣ i1 -YFP in cells expressing G␣ i -coupled GPCRs upon treatment with agonist. 5 The effect of receptor activation on the AGS4-G␣ i interaction led us to ask whether AGS4-G␣ i complexes were actually substrates for receptor-stimulated activation of G␣ i at the cell surface. We first examined the subcellular distribution of AGS4-YFP and G␣ i to determine whether their localization might overlap with that of a receptor present at the cell surface. Although AGS4 is primarily localized to the cytosol (supplemental Fig. S1) (1), co-expression with G␣ i resulted in a dramatic recruitment of AGS4-GFP to the cell cortex (supplemental Fig. S1), suggesting that AGS4-G␣ i complexes are at least within the same subcellular compartment as GPCRs, i.e. at the cell surface. G␣ i -mediated recruitment of AGS4-GFP to the cell cortex was not observed for AGS4-Q/A-GFP, nor in the context of the GTP hydrolysisdeficient G␣ i3 -Q204L variant (supplemental Fig. S1). As AGS4-RLuc-G␣ i -YFP BRET signals were dramatically reduced or absent in the context of the G␣ i -Q204L and AGS4-Q/A variants (Fig. 1), the co-localization of wild-type AGS4 and G␣ i suggest that their co-localization at the cell surface likely results from their interaction. We then measured agonist-induced changes in AGS4-RLuc and G␣ i1 -YFP distribution in crude membrane and cytosol fractions (supplemental Fig. S2). Receptor activation decreased the amount of AGS4-RLuc in the membrane fraction with a concomitant increase in the cytosol fraction as measured by both luminescence and immunoblotting (supplemental Fig.  S2A), whereas G␣ i1 -YFP distribution was unaltered (supplemental Fig. S2B). These data suggest that AGS4 and G␣ i physi- cally dissociate after receptor activation, with AGS4 moving into the cytosol and G␣ i remaining at the plasma membrane.
We then sought to determine whether AGS4 and G␣ i were actually forming a complex with receptors. Under basal conditions, specific BRET signals were not observed between AGS4-RLuc and ␣ 2A -AR-Venus (Fig. 3, A and B). However, when coexpressed with G␣ i , dramatic and specific BRET signals were observed (Fig. 3, A and B). 6 We did not observe significant alterations in either the basal or the agonist-induced changes in BRET observed between AGS4-RLuc and ␣ 2A -AR-Venus when either G␣ i1 or G␣ i3 was used (data not shown). BRET signals were not observed between AGS4-RLuc and ␣ 2A -AR-Venus in the context of the G␣ i -Q204L or G␣ i -G202T mutations or G␣ s (Fig. 3C), nor were they observed with the AGS4-Q/A-RLuc variant, which cannot bind G␣ i (data not shown), indicating that the AGS4-G␣ i interaction is required for the BRET signals observed between AGS4-RLuc and ␣ 2A -AR-Venus. In addition, the G␣ i -dependent AGS4-RLuc-␣ 2A -AR-Venus BRET signals were reduced by ϳ40 -50% upon treatment with the ␣ 2 -AR agonist UK14304 (Fig. 3, A and B). As was observed for the agonist-regulated BRET signals between AGS4-RLuc and G␣ i1 -YFP (Fig. 2), the reductions in G␣ i -dependent BRET signals between AGS4-RLuc and ␣ 2A -AR-Venus exhibited were dosedependent and occurred on a similar timescale (Fig. 3, D and E). The agonist-mediated reduction in AGS4-RLuc-␣ 2A -AR-Venus BRET was blocked by the antagonist rauwolscine (Fig. 3A) and by pertussis toxin pretreatment (Fig. 3F). Similar G␣ i -dependent and agonist-induced reductions in BRET signals were also observed between AGS4-RLuc and MOR-YFP but were not observed with AGS4-RLuc and ␤ 2 -AR-Venus, which is primarily a G␣ s -coupled receptor (Fig. 3G).
In the case of AGS4-RLuc-G␣ i -YFP BRET signals, receptor activation resulted in a reduction in BRET that could result either due to dissociation of AGS4-RLuc-Gi␣-YFP complexes or from a conformational rearrangement of the complex. The observation that receptor activation leads to a redistribution of AGS4 from the plasma membrane to the cytosol supports the first possibility. The reduction in agonist-modulated AGS4-RLuc-G␣ i -YFP BRET signal suggests that the agonist-regulated effect either occurs in a spatially restricted manner (e.g. at the plasma membrane) or is the result of second messengers. It is also possible that the agonist-induced reductions in BRET may arise from subunit exchange between the G␣ i -YFP complexed with AGS4-RLuc and wild-type, endogenous (i.e. untagged) G␣ i from G␣␤␥ heterotrimers or that "free" G␤␥ released from activated G␣␤␥ heterotrimers competes with AGS4-RLuc for G␣ i -YFP binding. Regarding the latter scenario, although data indicate that increased expression of G␤␥ subunits does indeed reduce AGS4-RLuc-G␣ i -YFP BRET (Fig. 1B), this is likely due to an overall reduction in the amount of AGS4-RLuc-G␣ i -YFP complexes being formed in the presence of excess G␤␥ rather than the release of free G␤␥ subunits occurring via receptor activation.
The data therefore suggest that AGS4-G␣ i complexes are directly coupled to and are regulated by a GPCR. The agonistinduced reduction in G␣ i -dependent BRET signals between AGS4-RLuc and ␣ 2A -AR-Venus (Fig. 3) and the agonist-induced reductions in AGS4-RLuc-G␣ i -YFP (Fig. 2) are consistent with nucleotide exchange occurring on G␣ i while complexed with AGS4-RLuc and suggest that once bound to GTP, the G␣ i subunit dissociates from AGS4-RLuc, resulting in the decrease in BRET signals in both cases. The pertussis toxin blockade of these agonist-regulated BRET signals suggests that the mechanism of activation of an AGS4-G␣ i complex is similar to that of receptor-mediated activation of G␣␤␥ heterotrimers and that GPCRs provide one mode of regulation of AGS4-G␣ i complexes. As AGS4 and other proteins containing more than one GPR motif can bind multiple G␣ i subunits simultaneously (7,8,19), these data have far reaching implications for G-protein signal processing and may provide cells with additional flexibility to modulate signal efficiency, specificity, duration, and location in ways previously unappreciated.