Modulation of Rap Activity by Direct Interaction of Gαo with Rap1 GTPase-activating Protein*

We used the yeast two-hybrid system to identify proteins that interact directly with Gαo. Mutant-activated Gαo was used as the bait to screen a cDNA library from chick dorsal root ganglion neurons. We found that Gαo interacted with several proteins including Gz-GTPase-activating protein (Gz-GAP), a new RGS protein (RGS-17), a novel protein of unknown function (IP6), and Rap1GAP. This study focuses on Rap1GAP, which selectively interacts with Gαoand Gαi but not with Gαs or Gαq. Rap1GAP interacts more avidly with the unactivated Gαo as compared with the mutant (Q205L)-activated Gαo. When expressed in HEK-293 cells, unactivated Gαo co-immunoprecipitates with the Rap1GAP. Expression of chick Rap1GAP in PC-12 cells inhibited activation of Rap1 by forskolin. When unactivated Gαo was expressed, the amount of activated Rap1 was greatly increased. This effect was not observed with the Q205L-Gαo. Expression of unactivated Gαo stimulated MAP-kinase (MAPK1/2) activity in a Rap1GAP-dependent manner. These results identify a novel function of Gαo, which in its resting state can sequester Rap1GAP thereby regulating Rap1 activity and consequently gating signal flow from Rap1 to MAPK1/2. Thus, activation of Go could modulate the Rap1 effects on a variety of cellular functions.

activates the cGMP phosphodiesterase (1), and G␣ i inhibits adenylyl cyclases (6) directly (7). However, direct effectors for G␣ o , an abundant G protein in the brain (8,9), have not yet been identified. G␣ o has been implicated in receptor-mediated inhibition of Ca 2ϩ channels in chick dorsal root ganglion neurons (10). Hence, it appeared feasible that this system could be used to identify proteins that directly interact with G␣ o . We used the yeast two-hybrid system to identify potential G␣ o effectors. For this purpose, we screened the chick dorsal root ganglion cDNA library with the mutationally (Q205L) activated form of G␣ o . In this article we present data indicating that the inactive form of G␣ o preferentially interacts with Rap1GAP 1 and thus regulates the activity of the small G protein Rap.

EXPERIMENTAL PROCEDURES
Materials-The cDNA synthesis system was from Life Technologies, Inc. Anti-G␣ o and anti-MAPK2 antibodies were from Santa Cruz Biotechnology, Anti-M2-FLAG antibody was from Sigma, anti-G␣ I-3/o antibody was from Upstate Biotechnology, Inc., and phospho-specific and total MAPK antibodies were from New England Biolabs. Most biochemicals were from Sigma, and cell culture supplies were from Life Technologies, Inc. All restriction enzymes were from New England Biolabs. Yeast culture media and amino acids were from CLONTECH. DNA plasmid preparation reagents and Effectene and Superfect transfection reagents were from Qiagen, Inc. ECL reagents were from Amersham Pharmacia Biotech. All other reagents were of the highest grade available.
Yeast Two-hybrid Screening-A directional oligo(dT)-primed cDNA library was constructed from 12-day embryonic chick dorsal root ganglion mRNA. cDNA was synthesized using a cDNA synthesis system and ligated into the GAL4 DNA-activation domain plasmid pPC86 using the SalI/NotI restriction sites (11). Plasmid DNA was isolated from the unamplified library using a Qiafilter Plasmid Maxi Kit. Q205L-G␣ o was cloned into the SalI/NotI restriction sites of the GAL4 DNA-binding domain plasmid pPC97-cycloheximide (11). The Q205L-G␣ o -BD plasmid and the library were co-transformed into the yeast strain MaV203, plated on selective media lacking leucine, tryptophan, and histidine and containing 25 mM 3-aminotriazole, and incubated at 30°C for 3 days. His ϩ colonies were then tested for ␤-galactosidase activity using a filter lift assay. Plasmid DNA was then isolated from positive yeast clones and reintroduced into MaV203 yeast expressing the GAL4 DNA-binding domain in-frame with either the wild type or Q205L-G␣ o cDNA. Positive clones were then sequenced, and BLAST analysis was performed using GenBank TM . The G␣ q , G␣ s , and G␣ i-2 cDNAs encoding both the wild type and the constitutively activated mutants were subcloned into the pPC97-cycloheximide plasmid.
CPRG Assay-The MaV203 yeast strain was co-transformed with the GAL4 DNA-binding domain and GAL4 DNA activation domain plasmids as indicated and incubated on selective media lacking tryptophan and leucine for 3 days at 30°C. Clones were grown for 24 h in selective liquid media and then inoculated into complete media and grown to an A 600 of 1.5. Yeast were lyzed with glass beads and incubated in a CPRG assay buffer (100 mM HEPES, pH 7.3, 154 mM NaCl, 4.5 mM L-aspartate, 1% bovine serum albumin, and 0.05% Tween 20) containing 2 mM CPRG (Roche Molecular Biochemicals) for 2-24 h. The reaction was stopped by the addition of ZnCl 2 to 1 mM, and the absorbance was measured at 574 nM.
Co-immunoprecipitations-HEK-293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum * This work was supported by National Institutes of Health Grants GM-54508 and DK-38671 (to R. I.) and CA-72971 (to P. J. S. S.). 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  and antibiotics. Cells were transfected using Effectene and harvested after 48 h in Buffer A (50 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 6 mM MgCl 2 , 10 g/ml aprotinin, 10 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride). One mg of whole cell lysate was pre-cleared for 1 h with protein G-agarose at 4°C. Immunoprecipitations were performed overnight with 5 g of M2-FLAG monoclonal antibody at 4°C followed by a 4-h incubation with 25 l of protein G-agarose. Samples were washed three times with phosphate-buffered saline (PBS) containing protease inhibitors and then resolved using SDS-PAGE. After transfer to nitrocellulose, the membranes were blocked in 5% nonfat dry milk, immunoblotted with anti-M2-FLAG (1 g/ml), anti-G␣ o (1:2000), or anti-G␣ i-3/o (1:2000) antibodies followed by either rabbit or mouse HRP-conjugated secondary antibody. Bands were visualized using the ECL detection method.
RalGDS Assay-PC-12 cells were grown in Dulbecco's modified Eagle's medium, 10% horse serum, 5% fetal calf serum, 0.5% glutamine and seeded at one million cells/10-cm dish. Cells were transfected with 10 g of FLAG-Rap1b (containing tandem FLAG epitopes at the N terminus of the Rap1b cDNA) in pcDNA3 with or without 10 g of wild type or the constitutively active versions of G␣ o or vector using Superfect per the manufacturer's instructions. Twenty-four h after transfection, the medium was replaced with low serum-containing media (0.1% horse serum) for 12 h. Cells were treated with 10 M forskolin/IBMX (3-isobutyl-1-methylxanthine) for 10 min, rinsed twice with ice-cold PBS, and lyzed in lysis buffer (10% glycerol, 1% Nonidet P-40, 50 mM Tris-HCl (pH 8.0), 200 mM NaCl, 2.5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 1 M leupeptin, 10 g/ml soybean trypsin inhibitor, 10 mM NaF, 0.1 M aprotinin, and 1 mM Na 3 VO 4 ). Activated Rap1b was isolated from cell lysates using a protocol adapted from Franke et al. (12). Lysates were clarified by centrifugation, and supernatants containing 1 mg of total protein incubated with 60 g of Gst-RalGDS-Rap binding domain (Gst-RalGDS was a gift form Dr. Bos, Utrecht University, The Netherlands to P. J. S. S.) pre-coupled to glutathione beads. After a 1-h incubation at 4°C, beads were pelleted and rinsed three times with lysis buffer, and protein was eluted from the beads using Laemmli buffer. Proteins were separated by electrophoresis on a 12% gel followed by transfer to a polyvinylidene difluoride membrane. Membranes were blocked in 5% milk for 1 h and probed with the anti-M2-FLAG monoclonal antibody followed by an HRP-conjugated anti-mouse monoclonal secondary antibody. Proteins were detected by enhanced chemiluminescence. The transfection efficiency of FLAG-Rap1b and G␣ o plasmids was evaluated using 20 g of cell lysate by immunoblotting using anti-FLAG and anti-G␣ o antibodies.
MAPK Immunoblots-PC-12 cells were transfected with 10 g of Flag-MAPK2 (containing tandem FLAG epitopes at the N terminus of MAPK2) in pcDNA3 with or without 10 g of wild type or constitutively active versions of G␣ o , Rap1GAP, or vector. Twenty-four h after transfection, the medium was replaced with low serum-containing medium (0.1% horse serum) for 12 h. Cells were stimulated with NGF (50 ng/ml) for 10 min, rinsed twice with ice-cold PBS, and lyzed in MAPK lysis buffer (10% glycerol, 1% Nonidet P-40, 50 mM Tris-HCl, pH 8.0, 137 mM NaCl, 2 mM EDTA, pH 8.0, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 10 g/ml aprotinin, 10 mM NaF, and 1 mM Na 3 VO 4 ). Five hundred g of whole cell lysate was immunoprecipitated for 2 h with 5 g of M2-FLAG monoclonal antibody pre-coupled to protein G-agarose. Samples were separated by gel electrophoresis followed by transfer to a polyvinylidene difluoride membrane. Membranes were blocked for 1 h in 5% milk and probed with anti-phospho-MAPK antibody (1:1000), which is specific for the active form of MAPK1/2, followed by mouse HRP-conjugated secondary antibody (1:10,000). Proteins were visualized using ECL detection. Controls for MAPK2 loading were performed using M2-FLAG monoclonal antibody.

RESULTS
Using the yeast two-hybrid system, we screened the chick DRG library with Q205L-G␣ o as the bait. Initial screens dorsal root ganglion positive interacting clones. Further analysis of the true positive clones allowed us to identify four interacting proteins in the initial screens. These proteins are listed in Table I. The sequences for all of these proteins have been submitted to GenBank TM , and the accession numbers are listed in Table I. The first protein identified was Gz-GAP, an RGS that has been characterized by other groups to selectively regulate the GTPase activity of G␣ z (13,14). We also identified the cDNA clone for a hitherto unrecognized RGS protein. We have named this protein RGS-17. Furthermore, we identified a pro-tein that does not appear to have functional homology with other known proteins. However, it has a region that is similar to clone called KIAA0514 in the human gene data base. These three clones at the present time have not been studied any further in our laboratory. The fourth G␣ o interacting protein we found was the chick homologue of the human Rap1GAP (15). This interaction turned out to be quite unusual and fascinating.
Because we were interested in G␣ o effectors, we hypothesized that such effectors would bind more avidly to the activated form G␣ o than to the inactive wild type form. To determine whether Rap1GAP interacted preferentially with Q205L-G␣ o , we used a yeast two-hybrid assay. Much to our surprise we found that Rap1GAP bound more avidly to the wild type-G␣ o than to Q205L-G␣ o (Fig. 1A). The interaction of Rap1GAP with G␣ o is specific, because no interactions were observed with either wild type or mutant activated G␣ s or G␣ q . The Rap1GAP did interact with G␣ i2 , but only a small difference could be detected between the active and inactive forms (Fig. 1B). We next determined whether G␣ o and Rap1GAP interacted within the context of a mammalian cell. For this determination, we tagged the Rap1GAP with the FLAG epitope at the N terminus and co-expressed it with wild type and Q205L-G␣ o in HEK-293 cells. Lysates from the cells were immunoprecipated with the anti-M2-FLAG antibody; the immunoprecipitate was resolved by SDS-polyacrylamide gel electrophoresis and blotted with an antibody against the carboxyl terminus of G␣ i-3 , which is specific for G␣ i-3 and G␣ o . Whole cell lysates from cells transfected with the empty FLAG vector did not show any G␣ o in the immunoprecipitates, even though both the wild type and activated G␣ o were being expressed (Fig. 2, left panels). However, when the Rap1GAP was co-expressed, the wild type G␣ o was more extensively imunoprecipitated than the Q205L-G␣ o . (Fig.  2, upper right panel). Rap1GAP also appears to interact with a native protein in HEK-293 cells, which is either G␣ o or G␣ i-3 . This finding is not surprising because it interacted with G␣ i-2 in the yeast two-hybrid assay. The levels of Rap1GAP in the immunoprecipitates are very similar (Fig. 2, middle right panel) as are the levels of expressed G␣ o and Q205L-G␣ o (Fig.  2, lower right panel).
We next determined whether the chick Rap1GAP has GTPase modulating activity. For this we transfected PC-12 cells with Rap1GAP and FLAG-tagged Rap1 (to allow for the examination of Rap1 in transfected cells). Cells were treated with or without 10 M forskolin to activate Rap1 (16). Whole cell lysates were prepared, incubated with Gst-RalGDS, which binds the GTP-bound form of Rap1 (12), and exposed to glutathione beads. Material adsorbed onto the beads was resolved by SDS-polyacrylamide gel electrophoresis and blotted with M2-FLAG antibody. When Rap1GAP cDNA was not used in the transfection, forskolin-dependent activation of Rap could be readily observed (Fig. 3, lanes 1 and 2) as previously demonstrated in these cells (16). However, when Rap1GAP was also used in the transfection no activation of Rap1 was observed (Fig. 3, lanes 3 and 4). We next determined whether G␣ o reg-

Interactions of G␣ o and Rap1GAP 21508
ulated the Rap1GAP modulation of Rap1 activity. Cells were transfected with wild type or Q205L-G␣ o with the Rap1GAP, and FLAG-tagged Rap1. Cells were treated with forskolin to activate Rap1. When G␣ o was used, even without forskolin, Rap1 activation was observed (Fig. 3, lane 5), presumably because G␣ o binds the transfected as well as the endogenous Rap1GAP inactivating it, thus allowing Rap1 activation. The addition of forskolin did not yield any further activation (Fig. 3,  lane 6), which suggests that endogenous Rap1GAP may be essential to regulate Rap1 activity. In contrast, when Q205L-G␣ o was used, activated Rap1 was inhibited under basal conditions and was not stimulated even by incubation with forskolin (Fig. 3, lanes 7 and 8). This finding may result from the inhibition of adenylyl cyclase by the activated form of G␣ o or some other unrecognized G␣ o signaling pathway. The experiments shown in Fig. 3 indicate that inactive but not activated G␣ o can facilitate the activation of Rap1 by sequestering Rap1GAP.
Because it is known that Rap1 regulates the activity of B-Raf and that B-Raf can regulate the activation state of MAPK1/2 (16), we next examined the effect of both wild type G␣ o and Q205L-G␣ o on MAPK signaling. For this determination, PC-12 cells were transfected with FLAG-tagged MAPK2 with and without Rap1GAP, G␣ o , or Q205L-G␣ o and then exposed to NGF to activate MAPK. When cells were treated with NGF, activation of MAPK2 was observed (Fig. 4, lanes 1 and 2) as had previously been demonstrated (17). When Rap1GAP was expressed, stimulation of MAPK2 by NGF was not observed (Fig.  4, lanes 7 and 8), indicating that the activation of MAPK2 is dependent on Rap1. When wild type G␣ o was expressed, there was a basal activation state of MAPK2 that could not be potentiated by NGF treatment (Fig. 4, lanes 3 and 4). However, when Q205L-G␣ o was expressed, MAPK2 activation by NGF was inhibited (Fig. 4, lanes 5 and 6), which is in agreement with the activation state of Rap1 (see Fig. 3). These data support the idea that unactivated G␣ o but not activated G␣ o is able to sequester Rap1GAP, thus regulating Rap1 signaling leading to an increase in the activated state of MAPK2.

DISCUSSION
The studies described here show that G␣ o in its unactivated state can selectively interact with Rap1GAP. The region of Rap1GAP involved in interaction with G␣ o is currently not known. However, the N terminus of Rap1GAP is highly conserved between the chicken and human sequence, and in fact all of the independent two-hybrid clones for chicken Rap1GAP contained the entire N terminus. This finding is intriguing because the N-terminal 35 amino acids of Rap1GAP are highly similar to regions conserved in two other G␣ interacting proteins: LGN, a G␣ i-2 interacting protein (18), and PCP-2, a guanine nucleotide exchange factor for G␣ o (19). This similarity suggests that the N terminus of Rap1GAP may be the region that interacts with G␣ o and that this conserved domain may be used by proteins to interact specifically with G␣ i family subunits.
This interaction between G␣ o and Rap1GAP results in the sequestration of Rap1GAP such that the levels of activated Rap1 increases. Thus, one might envisage that activation of G o would lead to the inhibition of Rap and consequently of Rapmediated signaling. Thus, G o like G i would be able to negatively modulate signaling by the cAMP pathway, but for G o this would occur at the level of Rap. In contrast to our current cannonical model of G protein regulation of intracellular signaling, we propose that G␣ o does so in a hitherto unrecognized manner. Rather than the activated form of G␣ o binding to and regulating an effector, it is the inactive form that binds an inhibitory protein (Rap1GAP). Activation of G␣ o would release Rap1GAP, which then would be free to inhibit the activity of Rap. Although this mechanism is quite the opposite of the manner in which other heterotrimeric G protein subunits regulate signal flow, it is not entirely implausible. First, G␣ o is the most abundant of the G protein ␣ subunits and is particularly abundant in the brain, as is Rap1GAP (15); hence, it is possible that there is enough G␣ o such that part of it can be used to sequester Rap1GAP. Second, G o is most often coupled to inhibitory receptors such as the ␣ 2 -adrenergic (20) and opiate receptors (21) in the brain. Activation of these receptors results in the inhibition of cAMP signaling, because cAMP is capable of activating Rap via both protein kinase A and the newly discovered exchange factors that directly bind cAMP (22,23). Interestingly, forskolin's activation of Rap1, but not that of G␣ o , was blocked by protein kinase A inhibitor (PKI; data not shown). G␣ o may be able to antagonize cAMP-dependent gene expression, especially signals routed through the Rap1 3 B-Raf 3 MAPK1/2 pathway (24), in neurons by interacting with Rap1GAP. The experiments shown in Fig. 4 support such a mechanism of regulation. Thus G o -coupled receptors could gate signal flow through the Rap1 to B-Raf pathway (25,26). The physiological significance remains to be determined.
This interaction of G␣ o with Rap1GAP is not likely to explain all of the biological actions of G␣ o . Activated G␣ o has been shown to trigger neurite outgrowth in neuronal cells (27). Although activated Rap1 can induce neurites in PC-12 cells (16), Rap1 activation was not detected in this study using activated G␣ o . Therefore, it does not appear feasible that the effects on Rap will explain how activated G␣ o stimulates neurite outgrowth nor its reported regulation of tyrosine kinases in chick dorsal root ganglion neurons (10). Other studies in our laboratory indicate that in NIH-3T3 cells, Q205L-G␣ o activates Stat-3 via Src. 2 The relevance of this pathway in neuronal cells is currently not known. The search for direct effectors regulated by activated G␣ o continues.