Receptor-independent Activators of Heterotrimeric G-protein Signaling Pathways*

Heterotrimeric G-protein signaling systems are activated via cell surface receptors possessing the seven-membrane span motif. Several observations suggest the existence of other modes of stimulus input to heterotrimeric G-proteins. As part of an overall effort to identify such proteins we developed a functional screen based upon the pheromone response pathway in Saccharomyces cerevisiae. We identified two mammalian proteins, AGS2 and AGS3 (activators of G-proteinsignaling), that activated the pheromone response pathway at the level of heterotrimeric G-proteins in the absence of a typical receptor. β-galactosidase reporter assays in yeast strains expressing different Gα subunits (Gpa1, Gsα, Giα2 (Gpa1(1–41)), Giα3(Gpa1(1–41)), Gα16(Gpa1(1–41))) indicated that AGS proteins selectively activated G-protein heterotrimers. AGS3 was only active in the Giα2 and Giα3genetic backgrounds, whereas AGS2 was active in each of the genetic backgrounds except Gpa1. In protein interaction studies, AGS2 selectively associated with Gβγ, whereas AGS3 bound Gα and exhibited a preference for GαGDP versus GαGTPγS. Subsequent studies indicated that the mechanisms of G-protein activation by AGS2 and AGS3 were distinct from that of a typical G-protein-coupled receptor. AGS proteins provide unexpected mechanisms for input to heterotrimeric G-protein signaling pathways. AGS2 and AGS3 may also serve as novel binding partners for Gα and Gβγ that allow the subunits to subserve functions that do not require initial heterotrimer formation.

The seven-membrane span hormone receptor coupled to het-erotrimeric G-proteins represents one of the most widely used systems for information transfer across the cell membrane. Signal processing via this system likely operates within the context of a signal transduction complex. Within such a signal transduction complex, there are likely accessory proteins (distinct from receptor, G-protein, and effectors) that participate in the formation of this complex and/or regulate signal transfer from receptor to G-protein. In addition, several reports suggest alternative modes of stimulus input to heterotrimeric G-proteins that do not require direct interaction of the G-protein with the seven-membrane span receptor itself. To identify such entities and to define putative components of such a signal transduction complex we initiated two broad experimental approaches (1)(2)(3)(4). One strategy focused on a functional readout involving G-protein activation and was based upon initial observations in our laboratory concerning the transfer of signal from R to G (3,4). This approach resulted in the partial purification and characterization of the NG10815 G-protein activator that directly increased GTP␥S binding to brain G-protein in the absence of a receptor. To extend this body of work, we developed an expression cloning system in Saccharomyces cerevisiae that was designed to detect mammalian activators of the pheromone response pathway in the absence of a G-proteincoupled receptor (5). The pheromone response pathway in S. cerevisiae incorporates a seven-membrane span receptor, a heterotrimeric G-protein, and a mitogen-activated protein kinase cascade that regulates mating behavior and growth (6). In this report, we present the identification and characterization of two proteins isolated in this functional screen. These proteins were termed AGS2 1 and AGS3 for activators of G-protein signaling.

EXPERIMENTAL PROCEDURES
Generation and Screening of cDNA Libraries-The NG108-15 cell line was propagated as described previously (4). mRNA was prepared from twenty 100-mm confluent plates of cells and was used to generate a cDNA library in the yeast expression vector pYES2 by standard procedures using reagents from Stratagene. The NG108-15 cDNA library was screened for activators of the pheromone response pathway as described (5). We used a his3 far1 yeast strain containing a genomic integration of the FUS1p-HIS3 reporter construct and lacking both the native pheromone response receptor Ste3 and Gpa1 G␣ subunit. A plasmid carrying a modified mammalian G i ␣ 2 subunit in which the amino terminus was replaced with the first 41 amino acids of Gpa1 was introduced into this strain. Essentially, the screen took advantage of the inducible expression of the library cDNAs by differential plating of transformants on selective medium.
cDNA Analysis-A rat brain cDNA library was screened with a 32 P-labeled 43-mer oligonucleotide (5Ј-ATAAGGCTGAAGAAATCCT-CATCAGGCATGGTAGGGCCAC-3Ј) derived from the AGS3 sequence cDNA isolated in the yeast screen. The longest brain AGS3 cDNA still contained a truncated reading frame, and the 5Ј-end was extended by 5Ј-RACE using Marathon-Ready cDNA (CLONTECH) from rat brain. Domain searches were performed by the GCG COMPARE/DOTPLOT PROFILESEARCH and MOTIFS commands via internet access to the simple modular architecture research tool (SMART) data bases (7). Multiple expectation maximization for motif elucidation version 2.2 (MEME), motif alignment and search tool (MAST), and profile searches were performed via web resources at the San Diego SuperComputer Center (8). Secondary structure analysis was performed with the GCG * This work was supported by the National Institutes of Health Grant RO1-NS24821 (to S. M. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  PEPPLOT, HELICALWHEEL, and MOMENT programs as well as by the PROBE program, the UCLA/DOE Fold Prediction Server, and the PredictProtein network server at EMBL-Heidelberg (9,10).
Immunoblotting and G-protein Interaction Assays-Protein interaction assays with purified G-protein subunits, GST fusion proteins, and immunoblotting were conducted as described previously (2). For protein interaction studies in crude cell lysates, confluent 100-mm dishes of DDT1-MF2 cells were lysed in 50 mM Tris-HCl, pH 7.4, 5 mM EDTA, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 60 M benzamidine, 40 M pepstatin, 1% Nonidet P-40 (500 l/dish) and were centrifuged at 14,000 ϫ g. 225 l of lysate containing GDP (30 M) or GTP␥S (30 M) were incubated with GST or GST⅐AGS3 (300 nM) for 30 min at 24°C or at 4°C for 12 h. 25 l of a 50% slurry of glutathione-Sepharose were added, and the incubation continued for 1 h. Samples were then centrifuged, and the matrix was washed with 2 ϫ 300 l of lysis buffer prior to solubilization and denaturing gel electrophoresis (2). For immunoprecipitation, antiserum generated against the carboxyl terminus of Gi3 (1 to 50 dilution of sera) was incubated with 250 l of precleared lysate for 16 h at 4°C, and immune complexes were isolated on Gammabind G-Sepharose. Bound proteins were detected by immunoblotting membrane transfers with specific antisera. The peptide C-VDLAG-SPEQEASGLPDPQQQYPPGAS was used for generation of AGS3 antisera in rabbits.

RESULTS AND DISCUSSION
To facilitate the identification of novel receptor-independent activators of heterotrimeric G-proteins, we developed an expression cloning system based upon the pheromone response pathway in S. cerevisiae. We used a yeast strain that lacked the pheromone receptor and contained a modified mammalian Gprotein (G i ␣ 2 ) in place of the yeast G-protein Gpa1 (5,11). The yeast strain was further modified to respond to activation of the pheromone response pathway with a readout of growth (5). As an initial source of such G-protein activators (4) we generated a NG108-15 cDNA library in the galactose inducible yeast expression vector pYES2. Three rounds of transformation/ screening with different libraries yielded three distinct cDNAs that promoted growth in a galactose-dependent manner (Fig. 1,  A and B). Epistasis experiments indicated that cDNA #34, which was the weakest of the three in the yeast screen, acted downstream of Ste5 (a component of the yeast mitogen-activated protein kinase cascade) in the pheromone response pathway, and this protein will be described elsewhere. cDNAs #37 and 53 did not function in the null Ste5 genetic background and were also inactive in yeast strains lacking Ste20 (yeast homolog of p21-activated kinases) or Ste4 (yeast homolog of G␤), indicating that these proteins activated the pheromone response pathway at or near the level of G-protein (Fig. 1B). Efforts were focused on cDNAs #37 and 53. Immunoblot analysis indicated that neither cDNA altered the levels of G␣ or G␤ subunits (data not shown). The selectivity of the two cDNAs for different G-protein heterotrimers was determined using yeast strains expressing Gpa1 (yeast G␣), G s ␣, G i ␣ 2(Gpa1(1-41)) , G i ␣ 3(Gpa1(1-41)), and G␣ 16(Gpa1(1-41)) . In ␤-galactosidase reporter assays, cDNA #37 was active in each of the genetic backgrounds except Gpa1, whereas cDNA #53 was only active in the G i ␣ 2 and G i ␣ 3 genetic backgrounds ( Fig. 2A). The preceding observations indicated that the activation of the pheromone response pathway by cDNAs #37 and 53 depended upon the presence of heterotrimeric G-proteins and the composition of subunit isoforms. cDNAs isolated via this expression cloning system were therefore named activators of G-protein signaling (AGS). cDNAs #37 and 53 were termed AGS2 and AGS3, respectively. AGS1, isolated from a human liver cDNA library, encodes a novel Ras-related protein (5). 2 AGS2 (770 nt) encoded a protein identical to mouse Tctex1, a FIG. 1. Activation of the pheromone response pathway by NG108-15 cDNA clones. A, three cDNA clones (#34, #37, and #53) were isolated from a screen of 1.1 ϫ 10 6 yeast transformants based upon their ability to promote growth in a galactose-dependent manner in the yeast expression cloning system (5). B, analysis of the cDNA clones in spot growth assays using yeast strains lacking STE20, STE5, or STE4. Control (noninduced, no selection), glucose medium lacking uracil and tryptophan. Noninduced selection, glucose medium lacking uracil/tryptophan/histidine and containing the histidine analog aminotriazole at 1 mM. Induced selection, galactose medium lacking uracil/tryptophan/ histidine and containing 1 mM aminotriazole. WT, wild type strain CY1316/1183. cDNA #34 was active in the absence of a functional STE5 and thus was not evaluated (black circles in key) in yeast strains lacking STE4 or STE20. For spot growth assays, ϳ2000 cells suspended in water were spotted on appropriate medium, and the plates were incubated at 30°C for 2 days prior to photography. Experiments in A and B were repeated three times with similar results .   FIG. 2. Functional and biochemical properties of AGS2 and  AGS3. A, bioactivity of cDNA #37 and 53 in yeast strains expressing different G␣ subunits. The Gpa1, G i ␣ 2 (Gpa1(1-41)) , G i ␣ 3 (Gpa1(1-41)) , and G␣ 16(Gpa1(1-41)) yeast strains expressed similar amounts of G-protein as determined by immunoblotting with Gpa1-specific antisera. Each series of experiments include data obtained in the absence of G␣ to indicate the signal obtained when the action of G␤␥ is not limited by reformation of heterotrimer. Data are presented as the mean Ϯ S.E. of three experiments with duplicate determinations. B, interaction of AGS2 and AGS3 with G-proteins. AGS2 and the bioactive AGS3 peptide (74 amino acids) beginning at the first in frame methionine were generated as GST fusion proteins. GST fusion proteins were incubated with G-protein subunits (G␣ ϭ 60 -80 nM, G␤␥ ϭ 40 nM, fusion protein ϭ ϳ300 nM, total volume ϭ 250 l) for 12 h at 4°C. GDP ϭ 10 M. GTP␥S ϭ 10 M plus 5 mM MgCl 2 . Proteins were then adsorbed to glutathione matrix and retained G-protein subunits identified by immunoblotting following gel electrophoresis. G i ␣ 2 was generated in Sf9 cells as an amino-terminal His6-tagged protein and was detected using anti-Xpress antisera (Invitrogen). Input, 20 l of the incubation mixture. The results presented are representative of four separate experiments. The GST⅐AGS3 fusion protein was functionally similar to the original AGS3 isolate in terms of its ability to promote growth. In contrast, the GST⅐AGS2 fusion protein did not promote growth in the yeast assay system, suggesting that the amino-terminal region of AGS2 has an important functional role. C, effect of G204A G i ␣ 2 substitution and RGS4 on the functionality of AGS proteins. AGS1 was discovered in a similar screen using a human liver cDNA library (5). WT, wild type strain CY1316/1183 containing G i ␣ 2 as described for the original yeast screen. Similar results were obtained in three experiments. light chain component of the cytoplasmic motor protein dynein and the flagellar dynein inner arm (12,13). Tctex1 also exists in the cell free of dynein where it may subserve as yet undefined roles in cellular signaling (14). Tctex1 physically associates with the G-protein-coupled receptor rhodopsin, the tyrosine kinase fyn, and a putative regulator of neurotransmitter release Doc2 (15)(16)(17). The functionality of AGS2 (Tctex1) in the yeast assay system also suggests a direct interaction with heterotrimeric G-proteins. This issue was addressed in protein interaction studies using a GST⅐AGS2 fusion protein and purified G-protein subunits. AGS2 did not interact with G i ␣ 2 but directly bound brain G␤␥ (Fig. 2B). The bioactivity associated with AGS2 in the yeast assay system suggests a regulatory role of heterotrimeric G-protein subunits in dynein function. Indeed, both G␣ and G␤␥ are implicated in various aspects of membrane trafficking, cytoskeletal dynamics, and vesicular transport (18 -22).
AGS3 consisted of 1500 nt and contained a truncated open reading frame at its 5Ј terminus with a methionine embedded in a Kozak's consensus sequence for translational initiation. The bioactivity/expression of this open reading frame (74 amino acids) was confirmed by mutational analysis and immunoblotting with specific peptide antisera (data not shown). The AGS3 peptide was generated as a GST fusion protein and evaluated for specific interactions with purified G-protein subunits (Fig.  2B). AGS3 specifically bound recombinant G i ␣ 2 , but it did not interact with purified brain G␤␥ (Fig. 2B). AGS3 preferred the GDP-bound conformation of G␣ and exhibited selectivity for G␣ subunits as observed in the functional assays described above (Fig. 2, A and B).
The protein interaction data and the functional data in yeast clearly indicated that AGS2 and AGS3 activated the pheromone response pathway by a process that involves heterotrimeric G-protein. As an initial approach to define their mechanism of action in the context of the cellular environment, we revisited the yeast system. We utilized yeast strains in which signal processing in the pheromone response pathway was compromised by replacement of the G i ␣ 2 subunit with a G204A site-directed mutant G i ␣ 2 or overexpression of the GTPase-activating protein RGS4 (23). G i ␣ 2 G204A behaves as a dominant negative G␣ subunit by virtue of its predicted low affinity for GTP, and it is incapable of supporting signal propagation by an activated receptor in our yeast system. 3 If AGS2 and AGS3 were activating G-proteins by a mechanism similar to a Gprotein-coupled receptor, then their action would be blocked in such genetic backgrounds. Surprisingly, this was not the case (Fig. 2C). In contrast, another AGS protein (AGS1) isolated in the same functional screen and described elsewhere (5), 2 was not active in the same genetic backgrounds thus serving as a positive internal control. AGS2 and AGS3 also did not alter GTP␥S binding to heterotrimeric brain G-protein or G i ␣ 2 (data not shown), which again contrasts with the effects of AGS1 and the NG108-15 G-protein activator (4, 5). 2 These data suggest that although AGS2 and AGS3 clearly activate the pheromone response pathway at the level of G-protein, this event does not require the generation of G␣GTP. By binding to G-protein subunits, AGS2 and AGS3 may inhibit heterotrimer formation or actively promote subunit dissociation "releasing" G␤␥. Such mechanisms of G-protein activation dramatically differ from that involving a typical G-protein-coupled receptor and might lead to selective activation of G␤␥-regulated effectors. Alternatively, AGS2 and AGS3 may position G-proteins within a signal transduction complex or regulate signal processing by compartmentalization. Increasing evidence indicates that G␣ and G␤␥ may exist independently of each other in the cell, and perhaps AGS2 and AGS3 serve as alternative functional binding partners for the G-protein subunits (20 -22).
A full-length rat AGS3 cDNA (650 amino acids; AF107723) was identified by screening a rat brain cDNA library and subsequent 5Ј-RACE (Fig. 3A). The full-length AGS3 protein, generated as an amino-terminal His-tagged protein in Sf9 cells, also interacts with G␣GDP as described earlier for the AGS3 subdomain isolated in the yeast screen. Endogenous full-length AGS3 and endogenous G i ␣ 3 apparently form complexes within the cell as determined in co-immunoprecipitation experiments FIG. 3. Consensus sequence for G-protein regulators. A, sequence of full-length rat AGS3. The shaded and overlined sequences represent the tetratricopeptide repeat motif and a repeated segment of amino acids (I-IV), respectively, as discussed in the text. The underlined amino acids indicate the fragment (cDNA #53) isolated in the original yeast screen. B, the four repeat domains in the carboxyl-terminal half of AGS3, human LGN, the predicted C. elegans protein, and related sequences identified by the motif search program MEME were aligned and used to define a consensus sequence. B, the AGS3 cDNA isolated in the yeast screen and shown to interact with G␣ subunits (Fig. 2B) was modified by site-directed mutagenesis to disrupt conserved residues within the GPR motif. The number of the amino acids altered in the mutant constructs (F8R, Q15A, and R23F) refers to their position in the sequence found in the left panel aligning the conserved domains. The GST⅐AGS3 fusion proteins were added to DDT1-MF2 cell lysates containing 30 M GDP and processed to determine interaction with endogenous mammalian G i ␣ 2 or G i ␣ 3 as described under "Experimental Procedures." C, 225 l of lysate containing nucleotide (30 M) were incubated with GST or GST⅐AGS3 (300 nM) for 30 min at 24°C. Protein complexes were isolated by glutathione affinity matrix or immunoprecipitation with G i ␣ 3 antisera, G i ␣ subunits were detected with the G i ␣ 3 antibody, and G␤ with an amino-terminal antibody. The images in B and C are representative of two to four experiments using different batches of fusion protein and cell extracts. Ab, antibody; Input, 10 l.
LGN was isolated as a truncated carboxyl-terminal fragment in a two-hybrid screen using G i ␣ 2 as "bait" (24). Analyis of human and mouse EST data bases indicated that AGS3 and LGN are not species homologs and likely encode two distinct members of a larger protein family defined by conserved structural motifs and functional properties. AGS3, LGN, and the predicted C. elegans protein actually possess two defined structural cassettes that essentially divide the protein in half. The amino-terminal half of the AGS3 consists of six tetratricopeptide repeats, which serve as protein interaction motifs and regulatory domains in various proteins. The tetratricopeptide repeats in AGS3 may function as a regulatory domain controlling the bioactivity of the carboxyl-terminal region or the trafficking of AGS3 and associated proteins within the cell.
The carboxyl-terminal half of AGS3 and the two related proteins contains four repeated sequences of 18 -19 amino acids that exhibit 80 -85% homology. One or two such repeat domains are also found in four proteins that influence the nucleotide-binding properties or GTPase activity of G-proteins (Fig. 3B). PcpL7 (PQ0109) (Pcp2) was isolated in a yeast twohybrid screen using G o ␣ as bait and may act as a guanine nucleotide exchange factor (25). RAP1GAP (P47736) is a GT-Pase-activating protein for the small G-protein RAP-1A and was also isolated in a two-hybrid screen using "activated" G o ␣ as bait (26). RGS12 (AF035151) and RGS14 (O087737) are members of the family of "regulators of G-protein signaling." The presence of such a repeated structural motif infers functionality, and we refer to the consensus sequence illustrated in Fig. 2B as a G-protein regulatory (GPR) motif. The GPR motif may be a signature for proteins that interact with G-proteins and regulate subunit interactions, nucleotide exchange, GTP hydrolysis, and/or the interaction of the G-protein with other entities involved in signal transduction.
Further analysis indicated that each GPR motif can exist as an amphipathic helix. The AGS3 cDNA isolated in the yeast screen actually contained one complete GPR motif. Introduction of mutations into this GPR motif that disrupt the predicted amphipathic helix eliminated interaction of AGS3 with G i ␣ 2 and G i ␣ 3 in crude cell extracts (Fig. 3B), and the same fusion proteins were inactive in the yeast assay system (data not shown). As observed for the interaction between AGS3 and purified recombinant G i ␣ 2 (Fig. 2B), AGS3 preferentially interacted with G␣GDP versus GTP␥S in the crude cell lysate (Fig.  3C). Despite the presence of GDP, which would stabilize the G-protein heterotrimer, the AGS3⅐G␣GDP complex from the mammalian cell lysate did not contain G␤␥. In contrast, G␤␥ subunits were readily detected when G␣GDP was isolated from the same cell extract by immunoprecipitation with a G␣ subunit antibody (Fig. 3C). These data support the hypothesis that AGS3 activates G-protein signaling by influencing subunit interactions. Alternatively, it is possible that AGS3 is selectively interacting with a population of G␣ in the cell that exists independent of G␤␥ and subserves unexpected functional roles. AGS proteins are indicative of a growing number of accessory proteins that influence signal propagation by heterotrimeric G-protein systems (3,4,19,25,26). Such entities may influence the population of activated G-protein within the cell independent of external stimuli or provide a cell-specific mechanism for signal amplification by acting in concert with G-protein-coupled receptors. Such proteins also provide a mechanism for signal input to heterotrimeric G-protein signaling systems that is distinct from that initiated by a seven-membrane span hormone receptor. By virtue of their distinct properties, AGS2 and AGS3 may also belong to a larger group of proteins that serve as binding partners for G␣ and G␤␥ allowing the subunits to subserve functions that do not require initial heterotrimer formation.