Dexras1/AGS-1 Inhibits Signal Transduction from the Gi-coupled Formyl Peptide Receptor to Erk-1/2 MAP Kinases*

Dexras1 is a novel GTP-binding protein (G protein) that was recently discovered on the basis of rapid mRNA up-regulation by glucocorticoids in murine AtT-20 corticotroph cells and in several primary tissues. The human homologue of Dexras1, termed activator of G protein signaling-1 (AGS-1), has been reported to stimulate signaling by Gi heterotrimeric G proteins independently of receptor activation. The effects of Dexras1/AGS-1 on receptor-initiated signaling by Gi have not been examined. Here we report that Dexras1 inhibits ligand-dependent signaling by the Gi-coupled N-formyl peptide receptor (FPR). Dexras1 and FPR were transiently co-expressed in both COS-7 and HEK-293 cells. Activation of FPR by ligand (N-formyl-methionine-leucine-phenylalanine (f-MLF)) caused phosphorylation of endogenous Erk-1/2 that was reduced by co-expression of Dexras1. Direct effects of Dexras1 on the activity of co-expressed, epitope-tagged Erk-2 (hemagglutinin (HA)-Erk-2) were measured by immune complex in vitro kinase assay. Expression of Dexras1 alone resulted in a 1.9- to 4.9-fold increase in HA-Erk-2 activity; expression of the unliganded FPR alone resulted in a 6.2- to 8.1-fold increase in HA-Erk-2 activity. Stimulation of FPR by f-MLF produced a further 8- to 10-fold increase in HA-Erk-2 activity over the basal (non-ligand-stimulated) state, and this ligand-dependent activity was attenuated at the time points of maximal activity by co-expression of Dexras1 (reduced 31 ± 6.8% in COS-7 at 10 min and 86 ± 9.2% in HEK-293 at 5 min, p < 0.01 for each). Expression of Dexras1 did not influence protein expression of FPR or Erk, suggesting that the inhibitory effects of Dexras1 reflect a functional alteration in the signaling cascade from FPR to Erk. Expression of Dexras1 had no effect on expression of Giα species, but significantly impaired pertussis toxin-catalyzed ADP-ribosylation of membrane-associated Giα. Expression of Dexras1 also significantly decreasedin vitro binding of GTPγS in f-MLF-stimulated membranes of cells co-transfected with FPR. These data suggest that Dexras1 inhibits signal transduction from FPR to Erk-1/2 through an effect that is very proximal to receptor-Gi coupling. While Dexras1 weakly activates Erk in the resting state, more potent effects are evident in the modulation of ligand-stimulated receptor signal transduction, where Dexras1 functions as an inhibitor rather than activator of the Erk mitogen-activated protein kinase signaling cascade.

Dexras1 is the prototypic member of a recently defined group of Ras-related intermediate molecular weight, basic GTP-binding proteins (G proteins). The group, which includes Dexras1, the activator of G protein signaling 1 (AGS-1) (1), the Ras homologue enriched in striatum (2), tumor endothelial marker 2 (TEM-2) (3), and Drosophila Dexras, is characterized by highly basic net isoelectric points and molecular weights intermediate between those of other Ras family members and the heterotrimeric G protein ␣ subunits (4). The increased molecular mass of these proteins is accounted for by a unique carboxyl terminus variable domain that is highly conserved within the group and bears no known structural motifs, except for a terminal consensus sequence for isoprenylation (CAAX box) (4). Dexras1 was first identified by differential display as a murine mRNA that is rapidly and transiently up-regulated by glucocorticoids in anterior pituitary, brain, and other tissues (5). We have reported that expression of Dexras1 in AtT-20 corticotroph cells results in the inhibition of stimulus-coupled peptide hormone secretion (4). Fang et al. recently identified the rat homologue of Dexras1 through its interaction with CAPON, a targeting protein for neuronal nitric oxide synthase (nNOS), implicating Dexras1 in N-methyl-D-aspartate receptor signaling in cortical neurons (6). The mechanism by which Dexras1 exerts its biological activities and the downstream signaling pathways governed by Dexras1 are not well characterized.
The recent discovery and characterization of a human homologue of Dexras1, AGS-1, establishes G i family heterotrimeric G proteins as potential signaling targets of Dexras1 (1,7). AGS-1 and Dexras1 share 97% amino acid identity, with conserved homology in the residues that vary (4). AGS-1 was identified functionally on the basis of its ability to activate ligand-independent signal transduction by a G i ␣ 2 /Gpa1 chimera in a yeast pheromone pathway-based genetic complementation screen (1). The yeast pheromone pathway (8) is analogous to the mammalian Erk-1/2 MAP 1 kinase pathway that is initiated by activation of many G i -coupled receptors, including the N-formyl peptide receptor (FPR), by their cognate ligand (9,10). The Erk-1/2 MAP kinase pathway mediates diverse signaling activities of receptor-tyrosine kinases, cytokine receptors and G protein-coupled receptors, including proliferation, differentiation, survival, and metabolism (9,11). Cismowski et al. used an indirect, luciferase-based transactivation assay to evaluate Dexras1 effects on this pathway in mammalian cells. AGS-1 expression had no effect on cAMP-response elementbinding protein or Jun basal transcriptional activities in COS-7 cells but produced a significant, 2.6-fold transactivation of Elk1 (7), an ETS domain transcription factor that lies downstream of Erk (12). These observations indicate that the effects of AGS-1 on yeast pheromone pathway signaling may translate to activation of the analogous Erk-1/2 MAP kinase pathway in mammalian cells. However, direct effects of Dexras1/AGS-1 on the phosphorylation state or kinase activity of Erk have not been demonstrated in mammalian cells. Furthermore, it is not known whether Dexras1/AGS-1 play a role in ligand-dependent signaling by G i -coupled receptors.
Activation of mammalian Erk MAP kinases by G i -coupled receptors is mediated through G␤␥, analogous to Ste4 and Ste18 in the yeast pheromone pathway (8). A role for G␤␥ in signaling by Dexras1 is supported by the observation that transactivation of Elk1 by Dexras1 is inhibited by overexpression of transducin-␣ (7), which acts as a scavenger of free G␤␥ (13). Therefore, analogous to receptor-mediated activation of G i , the interaction between Dexras1 and G i ␣ may promote release of G␤␥ and activation of G␤␥-mediated signaling pathways. Two observations further suggest that the Dexras1-G i ␣ interaction may functionally parallel the receptor-G i ␣ interaction. First, Dexras1 stimulates guanyl nucleotide exchange by G i ␣ to an extent similar to that reported for receptors (7); second, transactivation of Elk1 by Dexras1 is sensitive to pertussis toxin (7), which modifies an important site of coupling between G i ␣ and receptor (14).
Activation of G i ␣ by Dexras1 may require guanyl nucleotide binding by Dexras1, since a mutant form of Dexras1 reported to be deficient in guanyl nucleotide binding does not transactivate Elk1 (7). Therefore, we hypothesized that guanyl nucleotidebound Dexras1 (Dexras1⅐GTP or Dexras1⅐GDP) interacts with G i ␣ in a ligand-independent fashion to effect catalytic activities similar to those associated with ligand-dependent, receptorinitiated activation of G i ␣. According to this hypothesis, the release of G␤␥ from the G i heterotrimer initiated by Dexras1 would lead to ligand-independent activation of the Erk-1/2 MAP kinase pathway. Since Dexras1 potentially employs the same mechanism as receptor for activating G i ␣, we further hypothesized that Dexras1 might modulate G i -dependent signal transduction by a ligand-stimulated receptor. In view of data suggesting a physical interaction between Dexras1 and G i ␣ (1, 7), we considered that a Dexras1ϪG i ␣ interaction might result in cooperative or additive enhancement of G i signaling, resulting in increased downstream activity of Erk following receptor stimulation. Alternatively, a Dexras1ϪG i ␣ interaction might cause competitive or antagonistic interference with receptorϪG i ␣ signaling, resulting in inhibition of maximal Erk activity following receptor stimulation. The principle aim of this study was to distinguish between these potential modulatory effects of Dexras1 in the context of ligand-stimulated signaling by a G i -coupled receptor. Here we report that expression of Dexras1 increases Erk activity in COS-7 and HEK-293 cells but simultaneously attenuates further activation of Erk by ligand-stimulated FPR, acting through a mechanism that is proximal to receptorϪG i ␣ coupling.
Cell Culture and Transfection-Cells were maintained in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum and penicillin/ streptomycin. Cells were split 24 h prior to transfection so that 80% confluency was achieved at the time of transfection, and f-MLF treatments and collection of cell lysates were conducted 24 h following transfection. Transfection was performed using LipofectAMINE Plus reagent (Invitrogen) as recommended by the manufacturer. pcDNA3.1/His-C empty plasmid was used as a negative control for the Dexras1 expression plasmid. Cells for membrane preparations were obtained from cultures in multiple 100-mm plates, each transfected with 1 g of pcDNA3-FPR and 3 g of pcDNA3.1/His-Dexras1 or empty vector. Cells for Erk phosphorylation or kinase activity assays were cultured in 35-mm 6-well plates and transfected with: 0.33 g pcDNA3-FPR, 0.33 g pcDNA3.1/His-Dexras1, or empty vector and 0.33 g pSR␣-HA-Erk-2; pcDNA3.1 empty vector was substituted for pSR␣-HA-Erk-2 in experiments measuring the phosphorylation of endogenous Erk. Cells for luciferase reporter transactivation assays were cultured in 35-mm 6-well plates and transfected with 500 ng of reporter pFR-luc, 50 ng of expression plasmid for Elk1-GAL4dbd fusion protein, and 500 ng of either pcDNA-Dexras1 or control pcDNA3.1/His plasmids.
Elk1 Transactivation Assay-The PathDetect system (Stratagene, San Diego, CA) was used as recommended by the manufacturer to measure the transcriptional activity of an Elk1-GAL4dbd chimera against a luciferase reporter plasmid driven by the GAL4 upstream activation site (GAL4-UAS). Luciferase activity was measured in arbitrary luminescence units as previously described (16). Transfection of reporter plasmid with a plasmid expressing GAL4dbd lacking the Elk1 domain was used as a negative control alone and under various experimental conditions to establish background transcriptional activities that were subtracted from the final numbers reported. Luciferase activities for the Elk1-GAL4dbd chimera were 35-to 100-fold greater than baseline activities under various experimental conditions, demonstrating that regulatory effects were specific for the Elk1 transactivation domain.
Immune Complex Kinase Assays-HA-Erk-2 assays against the substrate myelin basic protein were performed in vitro with anti-HA immunoprecipitates, using the technique described by Graham et. al. (17). Phosphorylated myelin basic protein was isolated by separating the reactants on SDS-PAGE and analyzed by phosphorimaging on a Storm system (Molecular Dynamics, San Diego, CA). An aliquot of the reactants was further subjected to SDS-PAGE and Western blotting with a non-phosphoryl-specific Erk-1/2 antibody to determine the total quantity of HA-Erk-2 (New England Biolabs).
Membrane Preparation, GTP␥S Binding, and Pertussis Toxin Labeling-COS-7 cells were transfected as described above, and isolated membranes were prepared according to the method of Manning and co-workers. (18). GTP␥S binding to isolated membranes was assayed using a modification of the technique described by Seifert et. al. (19), differing only in that the binding buffer contained 5 M GDP and 1 nM [ 35 S]GTP␥S. In vitro pertussis toxin-mediated ADP-ribosylation of proteins in isolated membranes (25 g) was performed using a standard technique (20). Membrane proteins were solubilized after completion of the ADP-ribosylation reaction by boiling in Laemmli buffer and analyzed by SDS-PAGE; [ 32 P]ADP-ribosylation of G i ␣ species (ϳ41-kDa band) was imaged and quantitated by phosphorimaging.
Western Blotting-For analysis of activated, endogenous Erk-1/2, cells were treated, and lysates were prepared as for in vitro kinase assays; lysates were boiled with Laemmli buffer and analyzed by SDS-PAGE and Western blotting with phosphoryl-specific antibodies to Erk-1/2. For analysis of G i ␣ expression in membranes, 50 g of membrane protein was directly solubilized by boiling in Laemmli buffer and analyzed. A polyclonal antibody recognizing G i ␣ 1-3 was used to detect G i ␣ species. Secondary reagents and chemiluminescent detection substrate were provided by the Western Breeze system (Invitrogen) and used as recommended by the manufacturer. The relative intensities of bands on blots were quantitated by autoluminography using the Kodak 440cf digital imaging system (available through NEN Life Science Products).
Analysis of Cell Surface FPR Expression-Cell surface expression of FPR was quantitated by flow cytometric analysis of fluorescent ligand binding in COS-7 cells as described (21). Briefly, cells were transfected with expression plasmids for FPR and Dexras1 or empty control plas-mid in ratios comparable with those used for other assays. Cell samples were stained with 10 nM N-formyl-Nleu-Leu-Phe-Nleu-Tyr-Lys-fluorescein and analyzed by flow cytometry for the fraction of cells expressing the FPR. The mean channel fluorescence was determined for the population of FPR-expressing cells and compared with the population cotransfected with control plasmid.
Statistical Analysis-Quantitative data were analyzed using ANOVA, with post-hoc analysis using a paired t test and pairing assigned on the basis of replicate experiments. A p value of Ͻ 0.05 defined significant variation. All experiments were performed in duplicate or triplicate, and each independent experiment was repeated at least twice. Data are expressed as means Ϯ S.E.

RESULTS
Overexpression of AGS-1, the human homologue of Dexras1, has been reported by Cismowski et al. to activate Elk1, but not cAMP-response element-binding protein or Jun, as assessed by an indirect transactivation assay (PathDetect, Stratagene) (7). As shown in Fig. 1A, using the same assay, we observed a similar transactivation of Elk1 when mouse Dexras1 was overexpressed in either COS7 or HEK-293 cells. Dexras1-dependent activation of Elk1 was reduced by overnight pretreatment with pertussis toxin (89 Ϯ 4% reduction, p Ͻ 0.01) or by transient co-expression of the ␤-adrenergic receptor kinase carboxyl terminus peptide (ct-␤-ARK), a scavenger of free G␤␥ (65 Ϯ 4% reduction, p Ͻ 0.01) (22). We specifically tested for, but did not find, any difference in the protein expression level of Dexras1 in cells treated with pertussis toxin or co-expressing ct-␤-ARK (as detected by Western blotting, data not shown). The sensitivity to pertussis toxin is consistent with the reported mechanism of Dexras1/AGS-1 to activate the G i ␣ subunit. Release of free G␤␥ following ligand-stimulated receptor activation is believed to play an important role in the activation of Erk-1/2 MAP kinases (23). Thus, the inhibitory effect of ct-␤-ARK co-expression is consistent with a mechanism for Elk1 activation that involves release of free G␤␥ subunits as proposed by Cismowski et al. (7) The magnitude of Elk1 activation caused by expression of Dexras1 was modest (2.3 Ϯ 0.3-fold) and consistent with the 2.6-fold increase in Elk1 transactivation reported by Cismowski et. al. (7) Elk1 is a nuclear transcription factor whose phosphorylation and transactivation lie downstream of the Erk-1/2 MAP kinase signaling pathway in many cell types (12,24), suggesting that Dexras1 may utilize the Raf-1/MEK/Erk-1/2 signaling cascade in the transactivation of Elk1. Nevertheless, Erk-independent pathways for activation of Elk1 have also been described (25,26). To evaluate the effect of Dexras1 expression on more proximal signaling events in this pathway, we examined the ability of Dexras1 to influence Erk activity measured by an in vitro immune complex kinase assay utilizing co-transfected, hemagglutinin antigen-tagged Erk-2 (HA-Erk-2). As shown in Fig. 1B, lanes 1-5, co-transfection of COS-7 cells with an increasing quantity of expression plasmid for Dexras1 resulted in a dose-dependent increase in HA-Erk-2 activity against the myelin basic protein substrate, ranging from 1.9-fold (0.5 g of plasmid DNA) to 4.9-fold (2.0 g of plasmid DNA) over baseline. Overnight pretreatment of Dexras1-transfected cells with pertussis toxin or co-transfection of the ct-␤-ARK peptide resulted in reductions in HA-Erk-2 activity comparable in magnitude to the inhibitory effects of these agents on Elk1 transactivation (Fig. 1B, lanes 6 and 7).
To evaluate the effects of Dexras1 on ligand-stimulated signaling by a G i -coupled receptor, we co-expressed Dexras1 with the G i ␣ 2 /G i ␣ 3 -coupled (27) FPR in COS-7 and HEK-293 cells. Previous reports indicate that neither cell line expresses the endogenous FPR (28,29). We examined the effects of Dexras1 on FPR signaling before and after stimulation of cells with f-MLF, a peptide ligand of the FPR. As shown in Fig. 2 (lanes  1-3), stimulation with f-MLF caused a 6.3-fold maximal in-crease in phosphorylation of endogenous Erk-1/2 in each cell type as detected by phosphoryl-specific Western blotting. f-MLF had no effect on Erk-1/2 activation in cells transfected with the control plasmid instead of FPR (data not shown), which establishes the specificity of FPR-mediated Erk-1/2 activation and is consistent with the lack of endogenous expression of FPR by either cell line. The time course to maximal production of phospho-Erk-1/2 following addition of f-MLF was 5 and 10 min for HEK-293 cells and COS-7 cells, respectively (full time course data not shown). This time course of Erk-1/2 phosphorylation is comparable with that reported for other G protein-coupled receptors (22,30,31), including FPR (15). Expression of Dexras1 ( Fig. 2A, lanes 4 -6) resulted in a substantial decrease in the ligand-stimulated activation of Erk-1/2 in each cell line. Surprisingly, expression of Dexras1 did not affect baseline Erk-1/2 phosphorylation in this assay. This contrasts with the findings of up to 4.9-fold activation of HA-Erk-2 measured by the in vitro kinase assay. This discrepancy likely reflects methodological differences, perhaps related to the elimination of background activity in the HA-Erk-2 in vitro kinase FIG. 1. Expression of Dexras1 stimulates Elk1 transcriptional activity and Erk kinase activity. A, transactivation of an Elk1-GAL4dbd fusion protein with and without co-transfected Dexras1 was measured using the luciferase-based PathDetect system (Stratagene) in COS-7 cells. COS-7 cells were co-transfected with a GAL4-UAS-driven luciferase reporter, an expression plasmid for an Elk1-GAL4dbd fusion protein, and either an expression plasmid for Dexras1 or control plasmid (pcDNA3.1/His without insert). Dexras1 expression caused a 2.3 Ϯ 0.3% fold increase in Elk1 transcriptional activity. Co-expression of the carboxyl terminus fragment of the ␤-adrenergic receptor kinase (ct-␤-ARK) or overnight pretreatment with pertussis toxin (100 ng/ml) inhibited the effect of Dexras1 on Elk1 activity. * denotes p Ͻ 0.01 for increased Elk1 activity in Dexras1-transfected cells versus control plasmid-transfected cells; ** denotes p Ͻ 0.01 for decreased Elk1 activity in Dexras1-transfected cells treated as described versus untreated cells. B, transfection of COS-7 cells with increasing quantities of an expression plasmid for Dexras1, as indicated, resulted in a dose-dependent increase in the activity of co-transfected HA-Erk-2 (lanes 1-5). The effect of Dexras1 on HA-Erk-2 kinase activity was inhibited by pretreatment with pertussis toxin or co-expression of ct-␤-ARK (lanes 6 and 7). assay (Fig. 1B) that is otherwise contributed by non-transfected cells in the Western blotting assay (Fig. 2). Furthermore, the high constitutive activity of the non-ligand-stimulated FPR (discussed below) may have masked activity contributed by Dexras1 in the Western blotting assay.
As shown in Fig. 3A, lanes 1 and 3, and B, expression of FPR in the absence of ligand produced a significant increase in baseline HA-Erk-2 activity as assessed by the in vitro kinase assay. Ligand-independent signaling of G i ␣-coupled receptors, including FPR, to Erk-1/2 MAP kinases has been previously reported (19,32). In the case of FPR, ligand-independent signaling appears to be selectively sensitive to inhibition by the inverse agonist, cyclosporin H (19). The magnitude of HA-Erk-2 activity stimulated by FPR expression alone and independent of ligand was greater in magnitude to that stimulated by Dexras1 expression: 6.2-fold and 8.1-fold over baseline for COS-7 and HEK-293 cells, respectively (Fig. 3A, lanes 1 and 2,  and B). Interestingly, co-expression of Dexras1 with FPR but without ligand treatment produced little additional activation of HA-Erk-2 in HEK-293 cells beyond the activity resulting from expression of either Dexras1 or FPR alone (Fig. 3A, lanes  3 and 6, and B). In COS-7 cells, however, there was an approximately additive, statistically significant increase in activity under the same conditions, as shown in Fig. 3A (compare lanes  3 and 6 for COS-7) and B. Thus, Dexras1 may weakly augment the basal/constitutive signaling activity of FPR in a cell typespecific manner.
Following stimulation of FPR-transfected cells with f-MLF, HA-Erk-2 activity increased with a time course similar to that observed for Erk-1/2 phosphorylation (discussed above, Fig. 2). Peak HA-Erk-2 activities occurred at 5 and 10 min after stimulation in HEK-293 and COS-7 cells, respectively (see Fig. 3A,  lanes 3-5, and B). The maximal fold increases in activity over the resting state (8-to 10-fold) were comparable with those elicited by other G i -coupled receptors in these cell lines (22,31). HA-Erk-2 activities in cells co-transfected with Dexras1 were compared with control cells at the points of peak ligand-stimulated; as shown in Fig. 3A (compare lanes 3-5 and lanes 6 -8) and B, Dexras1 co-expression caused an 86 Ϯ 9.2% inhibition of f-MLF-stimulated HA-Erk-2 activity in HEK-293 cells (at 5 min) and a 31 Ϯ 6.8% inhibition in COS-7 cells (at 10 min, p Ͻ 0.01 for each cell line at those time points). The magnitude of this inhibitory effect on kinase activity was comparable with the inhibition of phosphorylation observed for endogenous Erk-1/2 in each cell line (discussed above, Fig. 2). In this series of experiments, the magnitude of the effect of Dexras1 expression on Erk activity was noticeably greater in HEK-293 cells than in COS-7 cells; this was apparent in effects of Dexras1 on both the stimulation of basal HA-Erk-2 activity and the inhibition of ligand-dependent signaling to HA-Erk-2. Since the magnitude of Dexras1 protein expression in detergent lysates or crude membrane fractions did not differ significantly between the two cell lines after correcting for total protein content (assayed by Western blotting, data not shown), we hypothesize that this difference in magnitude of effect on Erk may reflect differential regulation of Dexras1 activity or of a Dexras1ϪG i ␣ interaction through cell type-specific factors that have yet to be identified. Taken together, the data presented in Figs. 1-3 indicate that expression of Dexras1 activates a basal level of ligand-independent signaling by G i but simultaneously promotes a state that is refractory to further stimulation by ligand-stimulated receptor.
Studies were performed to determine whether the inhibitory effects of Dexras1 reflected alterations in the expression level of signaling pathway components (i.e. FPR, G i ␣, or HA-Erk-2). Cell surface expression of FPR was monitored by flow cytometric analysis of fluorescent ligand binding in COS-7 cells cotransfected with Dexras1 or control plasmids in ratios similar  1-3). Expression of Dexras1 inhibited activation of endogenous Erk-1/2 species following f-MLF stimulation (lanes 4 -6). to those described for the other assays. Dexras1 did not affect the mean expression of FPR in co-transfected cells (Fig. 4B). To test the effects of Dexras1 on HA-Erk-2 expression, anti-HA immunoprecipitates were subjected to Western blotting with a non-phosphoryl-specific antibody to Erk. Co-expression of HA-Erk-2 with FPR resulted in a slight decrease in HA-Erk-2 expression, but co-expression of Dexras1 had no additional effect (see Fig. 3A, Total Erk, and Fig. 4B). The membrane expression of endogenous G i ␣ species in COS-7 cells, as detected by Western blotting, was not affected by expression of Dexras-1 (Fig. 4A, top panel, and B). Together these data indicate that the stimulatory effects of Dexras1 on basal Erk activity and the inhibitory effects of Dexras1 on ligand-stimulated Erk activity cannot be accounted for by alterations in the expression level of FPR, G i ␣, or HA-Erk-2.
We also evaluated the effect of Dexras1 expression on ADPribosylation of G i ␣ species by pertussis toxin. In replicate experiments in isolated COS-7 cell membranes, shown in Fig. 4A,  bottom panel, and B, Dexras1 expression was associated with a 53 Ϯ 3.7% reduction in the ADP-ribosylation of a 41-kDa substrate corresponding to G i ␣ species. Pertussis toxin requires the heterotrimeric interaction of G i ␣ and G␤␥ to efficiently catalyze the ADP-ribosylation of G i ␣ (33). Activation of G icoupled receptors or inhibition of G␥ subunit posttranslational modifications have been shown to disrupt this interaction and reduce the ADP-ribosylation of G i ␣ in isolated membranes (33)(34)(35). The finding that Dexras1 expression causes a reduction in the ADP-ribosylation of G i ␣ without affecting the total amount of immunoreactive G i ␣ suggests that Dexras1 may disrupt the G i ␣-G␤␥ interaction in a manner similar to activated G i receptor, consistent with Dexras1-mediated activation of G i ␣ and release of G␤␥ as proposed by Cismowski, et al. (7) To evaluate the possibility that Dexras1 may interfere, competitively or otherwise, with coupling between receptor and G i ␣, we co-transfected Dexras1 or control plasmids with FPR in COS-7 cells and measured the effect of each on f-MLF-stimulated guanyl nucleotide exchange activity (GTP␥S binding) in isolated membranes. Ligand-stimulated GTP␥S binding in isolated membranes is considered one of the earliest events that can be measured following receptor-G protein coupling (36). In the case of FPR, ligand-stimulated GTP␥S binding in isolated membranes has been shown to be fully sensitive to pertussis toxin, consistent with observations that both G i ␣ 2 and G i ␣ 3 selectively interact with the ligand-stimulated FPR (27). As shown in Fig. 5, stimulation of membranes derived from cells that were not transfected with the FPR resulted in no change in GTP␥S binding (Fig. 5, crossbars and open triangles), consistent with the report that COS-7 cells do not express an endogenous FPR (28). In contrast, membranes expressing the FPR and stimulated with f-MLF (Fig. 5, solid squares) demonstrated significantly greater time-dependent GTP␥S binding over the zero point (p Ͻ 0.001 at all time points) with a saturable kinetic. It is noteworthy that the total (molar) GTP␥S binding differed from results obtained in a similar assay by Wenzel-Seifert et. al. (19) The discrepancy is likely explained by the difference in experimental conditions; in the study of Wenzel-Seifert et al., Sf9 cells were used (rather than COS-7 cells), and G protein subunits were overexpressed with the receptor (19), a condition that would be anticipated to provide far more receptor-coupled sites for specific binding of GTP␥S than in COS-7 cell membranes. Co-transfection of Dexras1 (Fig. 5, open squares) resulted in a substantial reduction in f-MLF-stimulated GTP␥S binding relative to membranes transfected with FPR only (Fig. 5, closed squares, p Ͻ 0.01 at all time points). Co-transfection of Dexras1 resulted in a reduction in f-MLF-stimulated GTP␥S binding comparable in mag-nitude to that seen in membranes derived from pertussis toxintreated cells. At two time points in the binding assay (15 and 30 min), the magnitude of response in Dexras1-transfected membranes (Fig. 5, open squares) was significantly greater than in control membranes derived from pertussis toxin-treated cells (Fig. 5, closed circles, p Ͻ 0.05 at 15 min, and p Ͻ 0.01 at 30 min); however, the response did not differ statistically at other time points in the series (7.5, 60, 120, and 180 min). The finding that expression of Dexras1 inhibits FPR-stimulated GTP␥S binding in isolated membranes indicates that Dexras1 may be FIG. 4. Expression of Dexras1 does not alter expression of signaling pathway components but inhibits ADP-ribosylation of G i ␣ species by pertussis toxin in isolated membranes. A, COS-7 cells were transfected under conditions that yielded maximal expression of Dexras1, and membranes were isolated by hypotonic lysis and differential centrifugation as described under "Experimental Procedures." G i ␣ was detected with an apparent mobility of 41 kDa by Western blotting with a polyclonal antibody that recognizes G i ␣ 1-3 (arrow labeled G i ␣). Enzymatic ADP-ribosylation of membrane proteins was performed in the presence of the pertussis toxin-free catalytic subunit and 32 P-labeled ␤-NAD, precisely as described by Carty (20) and analyzed by SDS-PAGE and phosphorimaging for the incorporation of radiolabel into ϳ41-kDa G i ␣ species (arrow labeled 32 P-ADPR-G i ␣). B, expression of signaling pathway components in Dexras1-transfected cells relative to control (Percent Control) and comparison with ADPribosylation of G i ␣ in membranes. Quantitation of cell surface FPR expression (open bar) was performed by fluorescence-activated cell sorter analysis of fluorescent ligand binding to the surface of FPRtransfected COS-7 cells as described under "Experimental Procedures." Total HA-Erk-2 (solid bar) was detected by Western blotting in detergent lysates of unstimulated COS-7 cells as in Fig. 3A and quantitated by chemiluminescence using a Kodak 440cf digital imaging system. G i ␣ species (gray bar) in equal quantities of isolated COS-7 membranes were detected and quantitated by Western blotting as described above. Quantitation of pertussis toxin-catalyzed ADP-ribosylation was performed by volume integration using a phosphorimaging system. Data represent the mean of two replicate experiments for each measurement Ϯ S.E.; * denotes p Ͻ 0.01 for decreased ADP-ribosylation relative to control.
acting at a point that is relatively proximal to receptor-G i coupling.

DISCUSSION
The present study confirms that Dexras1 is capable of exerting important regulatory effects on G i -mediated signal transduction to the Erk-1/2 MAP kinase cascade. In particular, findings in the present study suggest that the effects of Dexras1 on G i -mediated signaling are not monotonic but rather depend upon the signaling context and presumably activation status of G i ␣. We have examined the effects of Dexras1 expression in three distinct settings: (i) baseline resting conditions; (ii) receptor-mediated but ligand-independent activation of G i , reflective of conditions under which inverse agonists are operationally defined (32); and (iii) classical ligand-dependent activation of a G i -coupled receptor, in this case FPR. If we assume for purposes of discussion that over-expression of Dexras1 reflects the intrinsic activities of the endogenous protein, then it appears that under baseline conditions, Dexras1 favors the GTP-bound state of G i ␣ and the dissociation of G␤␥. Under conditions of ligand-dependent receptor-mediated activation, Dexras1 has opposing effects, in which Dexras1 interferes with receptor activation of G i . Intermediate effects are observed under conditions of ligand-independent receptor-mediated activation The finding that Dexras1 can stimulate Erk activity but simultaneously inhibit further activation of Erk by a G i -coupled receptor raises questions about the mechanism of Dexras1 action. Receptor-independent transactivation of Elk1 by Dexras1/AGS-1 appears to occur through a mechanism similar to that employed by G i -coupled receptors (7). The rate of guanyl nucleotide exchange stimulated by Dexras1 in vitro is comparable with rates described for ligand-stimulated, G i -coupled receptors (7). Further, activation of Elk1 by Dexras1 is inhibited by pertussis toxin (7), which catalyzes the carboxyl-terminal ADP-ribosylation of most G i / o family ␣ subunits. Because ADP-ribosylation is believed to inhibit G protein signaling by preventing the interaction of receptor and G␣ subunit (26), this observation suggests that Dexras1 and receptor may interact with the same region of G i ␣. Transactivation of Elk1 is also inhibited by co-expression of a scavenger of free G␤␥ (7), indicating that Dexras1 may signal to Elk1 via G␤␥-dependent activation of the Ras/Raf/MEK/Erk cascade (9). Our finding that expression of Dexras1 inhibits the ADP-ribosylation of G i ␣ provides additional evidence that Dexras1 promotes the release of G␤␥ from activated G i ␣. Together these observations suggest that Dexras1/AGS-1 and intracellular receptor domains may interact with G i ␣ through the same structural elements of the ␣ subunit and through a similar molecular mechanism. Based on these findings, we hypothesize that the ability of Dexras1 to selectively inhibit receptor-mediated G i signal transduction and render G i ␣ species resistant to pertussis toxin-mediated ADP-ribosylation could reflect a dominant effect of Dexras1 on G i the activation status of G i ␣ that results in a depletion of the pool of heterotrimeric G i ␣Ϫ␤␥ available to the receptor for transmitting ligand-dependent signals.
Evidence presented by Cismowski et al. (1,7) and in this report suggests that Dexras1 modulates G i -mediated signal transduction at proximal signaling events, probably including a direct effect of Dexras1 on the guanyl nucleotide binding state of G i ␣. In vitro studies have confirmed that activation of G i ␣ by AGS-1 may involve a direct binding interaction between the two proteins that increases the guanyl nucleotide exchange rate of the ␣ subunit and leads to the formation of G i ␣⅐GTP (7). This direct interaction between a Ras family G protein and heterotrimeric G protein represents a novel paradigm for signal transduction and suggests that Dexras1/AGS-1 may be functionally classified as a guanyl nucleotide exchange factor (GEF) for G i ␣ proteins. An in vivo G i ␣-GEF function for Dexras1 is less clearly established, and direct in vivo interactions between Dexras1/AGS-1 and G i ␣ have not been reported. The finding that activation of Erk by Dexras1 in a transfected cell system is inhibited by pertussis toxin is consistent with a potential G i ␣-GEF activity of Dexras1 in vivo. However, pertussis toxin-sensitive signaling from G i -coupled receptors to Erk appears to depend on the release of G␤␥ (9), and regulation of a direct target of G i ␣⅐GTP by Dexras1 (e.g. adenylate cyclase) has yet to be established. It will be important in future studies to determine whether Dexras1 also activates the ␣ subunit-dependent arm of G i signal transduction in vivo.
The novel hypothesis developed by Cismowksi et al., in which a Ras family G protein directly modulates the activation of a heterotrimeric G protein subunit (7), has important physiological implications for the many systems that utilize G i -coupled receptors. It will be important to discern whether Dexras1 behaves in a similar manner with other G i -coupled receptors and to define the arc of upstream signals and downstream effectors that describe the Dexras1 signaling pathways. Our findings establish an inhibitory activity of Dexras1/AGS-1 on ligand-dependent signaling of a G i -coupled receptor that is likely to be physiologically important and mechanistically related to ligand-independent activation of G i ␣ for which this founding member of the family of intermediate molecular weight, basic G proteins was named.