RA-GEF-1, a guanine nucleotide exchange factor for Rap1, is activated by translocation induced by association with Rap1*GTP and enhances Rap1-dependent B-Raf activation.

We previously identified RA-GEF-1, a novel guanine nucleotide exchange factor (GEF) for Rap1 with the ability to associate with Rap1.GTP at its Ras/Rap1-associating (RA) domain. Because it possesses a PSD-95/DlgA/ZO-1 (PDZ) domain, it was also named PDZ-GEF. In this report, we have examined the role of the RA domain of this protein in Rap1-mediated cellular responses. A mutant of RA-GEF-1 (RA-GEF-1DeltaRA) carrying a 21-residue deletion at its RA domain fully retains the in vitro GEF activity toward Rap1 but completely loses the Rap1 binding activity. In contrast, RA-GEF-1DeltaRA, expressed in COS-7 cells, exhibits a 3-fold reduction in its in vivo GEF activity toward Rap1 compared with wild-type RA-GEF-1 as examined by the Rap1 pull-down assay. Correspondingly, when coexpressed with wild-type Rap1, RA-GEF-1DeltaRA is unable to further activate B-Raf, whereas RA-GEF-1 stimulates B-Raf as efficiently as activated Rap1. Consistent with these observations, coexpression of activated Rap1 induces translocation of RA-GEF-1, which is otherwise located in the cytoplasm, to the perinuclear compartment, where Rap1 is also predominantly localized. This localization almost coincides with that of the Golgi apparatus, which was detected by anti-trans-Golgi-network 38 antibody. RA-GEF-1DeltaRA fails to show the translocation. These results indicate that RA-GEF-1 defines a novel category of GEF that is translocated to a particular subcellular compartment by association with the GTP-bound form of a small GTPase and catalyzes activation of the GDP-bound form present in the compartment, thereby causing an amplification of cellular responses induced by the small GTPase.

Ras family small GTPases have been implicated as a molecular switch that directs cell proliferation and differentiation by cycling between GTP-bound and GDP-bound forms (1,2). The GTP-bound form is active in that it directly binds to and activates specific effector molecules (3). The effector region of Ras family GTPases (amino acids 32-40 in human Ha-Ras) is involved in the interaction with effectors (1). The Ras-binding domain of the serine/threonine kinase Raf-1, one of the best characterized Ras effectors, interacts directly with the effector region of Ras in a GTP-dependent manner. On the other hand, the RA 1 domain, which is also responsible for binding to Ras, was identified in several Ras effectors such as RalGDS and AF-6/Afadin (4). The tertiary structure of the Ras-binding domain of Raf-1 is similar to that of the RA domain of RalGDS, although no obvious homology was found in their amino acid sequences (5)(6)(7).
The Ras family consists of ϳ20 members, including Ha-Ras, Ki-Ras, N-Ras, Rap1A, Rap1B, Rap2A, Rap2B, R-Ras, R-Ras2/ TC21, R-Ras3/M-Ras, RalA, and RalB. Among them, Rap1 was originally characterized as an antagonist of Ki-Ras-induced transformation and is thus termed Krev-1 as well (8). The ability of Rap1 to block transformation is likely ascribed to its competitive binding to Ras effectors because Rap1 shares the effector region with Ras and in fact associates with a subset of Ras effectors such as Raf-1 without stimulating their activities. This property of Rap1 is attributable to its greatly enhanced interaction with the cysteine-rich domain, a second Ras-binding site, of Raf-1 (9,10). In mammalian cells, including fibroblasts, platelets, T and B lymphocytes, and neutrophils, Rap1 is activated in response to a diverse array of extracellular stimuli. Interleukin-2 gene transcription in T cells and insulininduced mitogen-activated protein kinase activation in Chinese hamster ovary cells, for instance, are presumed to be regulated by both positive and negative actions on Raf-1 exerted by Ras and Rap1, respectively (2). However, Rap1 is rapidly activated after various stimulations without affecting the Ras signaling pathway (11,12). Therefore, it is feasible that Rap1 also exerts its own physiological function other than the modulation of Ras-dependent pathways. In nerve growth factor-triggered signaling in PC12 pheochromocytoma cells, Rap1 is reported to be involved in B-Raf activation, leading to the sustained activation of extracellular signal-regulated kinases that is required for neuronal differentiation (13,14). Rap1-dependent activation of B-Raf, but not Raf-1, was observed in COS-7 cells as well (10). Furthermore, the role of Rap1 in integrin-mediated leukocyte adhesion has recently been delineated. An active form of Rap1 potently induced the activation of integrins and subsequent cell aggregation (15,16), whereas a dominant-negative form of Rap1 and GTPase-activating proteins for Rap1, Rap-GAP, and SPA-1 inhibited cell adhesion triggered by extracellular stimulations including T-cell receptor or CD31 ligation (15)(16)(17). Lipopolysaccharide-induced activation of integrins in macrophages also requires Rap1 (18).
Mechanisms underlying the regulation of Rap1 remain largely unknown. Like Ras, Rap1 is believed to be activated by specific GEFs such as smgGDS (19), C3G (20), Epac/cAMP-GEF (21,22) and CalDAGGEFI (23). Recently, we identified a novel type of Rap1 GEF in humans (Hs-RA-GEF; referred to from here on as RA-GEF-1) and Caenorhabditis elegans (Ce-RA-GEF) (24). Other groups also reported a molecule identical to RA-GEF-1 and designated it PDZ-GEF1 (25), nRapGEP (26), and CNrasGEF (27). RA-GEF-1 contains the cNMP-binding, Ras exchanger motif and PDZ and RA domains as well as the GEF catalytic domain. The RA domain of RA-GEF-1 associates directly with the GTP-bound form of Rap1, whereas it exhibits no detectable binding to Ha-Ras. On the other hand, RA-GEF-1 shows GEF activity toward Rap1 and Rap2, but not Ha-Ras. However, regulatory mechanisms of GEF activity remained obscure. We and others detected no specific cAMP/cGMP binding to the cNMP-binding domain of RA-GEF-1 (24 -26), although Pham et al. (27) reported cAMP binding to this domain and subsequent stimulation of Ras GEF activity.
Here we investigated the role of the RA domain in the regulation of RA-GEF-1. We show that a deletion mutation within the RA domain that virtually abolishes Rap1 binding significantly diminished GEF activity in the cell, whereas it did not affect GEF activity in vitro. Additionally, Rap1-dependent translocation of RA-GEF-1 to the perinuclear compartment was observed, which was totally abolished by the RA domain mutation. Hence, GEF activity of RA-GEF-1 in vivo is likely to be enhanced through the RA domain-mediated translocation to the perinuclear region, where Rap1 exerts its function.
In Vitro Rap1A Binding Assay-The post-translationally modified form of Rap1A was purified from Spodoptera frugiperda Sf9 cells infected with baculovirus overexpressing Rap1A as described previously (9). The in vitro binding assay was carried out by incubating 20 l of amylose resin carrying MBP-RA WT or MBP-RA MUT with guanosine 5Ј-O-(3-thiotriphosphate)-or guanosine 5Ј-O-(2-thiodiphosphate)loaded Rap1A in a total volume of 100 l of buffer A (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 5 mM MgCl 2 , and 0.1% Lubrol PX). After incubation at 4°C for 2 h, resin was washed, and bound proteins were eluted with buffer A containing 10 mM maltose and subjected to SDS-polyacrylamide gel electrophoresis. Immunoblot detection was performed using anti-Rap1A antibody (sc-65) and enhanced chemiluminescence reagents (Roche Molecular Biochemicals).
In Vitro GEF Assay-FLAG-RA-GEF-1 and FLAG-RA-GEF-1⌬RA were expressed in Sf9 cells and affinity-purified with agarose resin conjugated with the anti-FLAG antibody M2 (Sigma). GEF assays were performed as described previously (24). Briefly, 2 pmol of Rap1A loaded with [ 3 H]GDP (5,000 cpm/pmol) was incubated with 1 g of FLAG-RA-GEF-1 or FLAG-RA-GEF-1⌬RA in 50 l of reaction buffer containing 20 mM Tris-HCl, pH 7.4, 3 mM MgCl 2 , 50 mM NaCl, 10 mM 2-mercaptoethanol, 5% glycerol, 5 mg/ml bovine serum albumin, and 3 mM unlabeled GTP for the indicated periods. The reaction was terminated by the addition of 2 ml of ice-cold stop buffer containing 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 5 mM MgCl 2 , and the reaction mixture was subjected to filtration through a nitrocellulose membrane (0.22-m pore size). After washing with stop buffer, the membrane-trapped radioactivity was measured by liquid scintillation counting.
In Vitro Association with cAMP and cGMP-FLAG-RA-GEF-1 was purified from Sf9 cells as described above. GST-fused protein kinase A regulatory subunit I␣ (GST-PKA-RI␣) was prepared as described previously (24). The cAMP binding assay was performed essentially as described previously (24), with minor modifications. FLAG-RA-GEF-1 (1 g) immobilized on anti-FLAG M2 resin or GST-PKA-RI␣ (0.5 g) immobilized on glutathione-agarose resin was incubated in a 100-l reaction mixture containing 10 mM potassium phosphate, pH 6.8, 150 mM NaCl, 1 mM EDTA, 100 g/ml bovine serum albumin, 25 mM ␤-mercaptoethanol, and 500 nM [2, H]cAMP (10,000 cpm/pmol) at 25°C for 90 min with gentle shaking. After extensive washing, bound proteins were eluted with 0.2% SDS and counted for 3 H label. The cGMP binding assay was carried out in a similar manner, except that [8-3 H]cGMP (5,000 cpm/pmol) replaced cAMP.

RA Domain Mutation That Abolishes the Association with
Rap1A-RA-GEF-1 acts as a GEF for Rap1 and Rap2 (24 -27), and its RA domain efficiently associates with Rap1A⅐GTP (24) but not with Rap2A⅐GTP (data not shown) in vitro. However, the role of the RA domain in the regulation of GEF activity remained unknown. To clarify this point, we first tried to identify a mutation in the RA domain that abolishes the binding activity. Arg-20 of RalGDS is critical for the association between RalGDS-RID and Rap1, as indicated by the finding that the R20A mutation of RalGDS completely eliminated the binding activity (6). Although this Arg residue is conserved in the RA domain of RA-GEF-1 (Arg-611), RA-GEF-1 carrying the R611A mutation still retained a residual binding activity toward Rap1A⅐GTP (data not shown). Subsequently, we deleted 21 N-terminal amino acids (amino acids 606 -626) of the RA domain by oligonucleotide-directed mutagenesis (Fig. 1A), and the association of MBP-RA WT and MBP-RA MUT with GDPbound and GTP-bound forms of Rap1A was examined (Fig. 1B). Whereas MBP-RA WT associated with Rap1A in vitro in a GTPdependent fashion, as we described previously (24), MBP-RA MUT exhibited no detectable binding to Rap1A.
Effect of the RA Domain Mutation on GEF Activity of RA-GEF-1 in Vitro-To examine the effect of the RA domain mutation on GEF activity, we assayed the GEF activities of fulllength wild-type RA-GEF-1 and its RA domain mutant (RA-GEF-1⌬RA). Both proteins were expressed with a FLAG tag in Sf9 cells and purified to near homogeneity by using anti-FLAG M2 resin. As shown in Fig. 2A, both of these proteins exhibited virtually the same enzymatic activity to stimulate GDP release from Rap1A. Therefore, the internal deletion in the RA domain does not interfere with the intrinsic activity of the GEF domain. It has been reported that binding of a ligand to the regulatory domain of certain GEFs, such as Epac/cAMP-GEF (21,22) and Asef (28) Effect of the RA Domain Mutation on GEF Activity of RA-GEF-1 in Vivo-GEF activities of RA-GEF-1 and RA-GEF-1⌬RA in vivo were examined by pull-down assays for Rap1A⅐GTP. RA-GEF-1 and RA-GEF-1⌬RA were expressed as FLAG-tagged proteins in COS-7 cells in combination with HAtagged wild-type Rap1A (HA-Rap1A WT ). Cell lysates prepared from transfectants were incubated with immobilized GST-Ral-GDS-RID, and bound HA-Rap1A WT ⅐GTP was quantitated by immunoblotting (Fig. 2B). Coexpression of FLAG-RA-GEF-1 caused a 6-fold increase in the amounts of the GTP-bound form of HA-Rap1A WT , which were almost equivalent to those of constitutively active HA-Rap1A (HA-Rap1A V12 ). In contrast, the GTP-bound Rap1A level in FLAG-RA-GEF-1⌬RA-expressing cells was ϳ3-fold lower than that in cells expressing FLAG-RA-GEF-1.
Effect of the RA Domain Mutation on B-Raf Kinase Activation in Response to RA-GEF-1-We next examined B-Raf activation following expression of RA-GEF-1 and its RA domain mutant on the basis of previous reports that Rap1⅐GTP activates B-Raf (10,13,14). To this end, FLAG-RA-GEF-1 or FLAG-RA-GEF-1⌬RA was expressed in combination with FLAG-B-Raf and either HA-Rap1A WT or HA-Rap1A V12 in COS-7 cells. FLAG-B-Raf was immunoprecipitated from total cellular extracts and subjected to kinase assays as described under "Experimental Procedures." As shown in Fig. 3, expression of HA-Rap1A WT resulted in only a 2-fold increase in kinase activity of FLAG-B-Raf. As expected, FLAG-B-Raf coexpressed with HA-Rap1A V12 showed significantly increased activity. When coexpressed with FLAG-RA-GEF-1 and HA-Rap1A WT , FLAG-B-Raf also exhibited increased kinase activity equivalent to that induced by HA-Rap1A V12 . In marked contrast, when coexpressed with FLAG-RA-GEF-1⌬RA and HA-Rap1A WT , the activity of FLAG-B-Raf was similar to that induced by HA-Rap1A WT alone. These results parallel the lower level of HA-Rap1A⅐GTP within the cell expressing FLAG-RA-GEF-1⌬RA.
RA-GEF-1 Colocalizes with Rap1A in the Perinuclear Region-Rap1 was reported to localize predominantly in perinuclear regions, such as the Golgi apparatus and cytoplasmic vesicles (29 -32). To confirm this, colocalization of EGFP-Rap1A V12 with trans-Golgi-network 38, which localizes primarily to the trans-Golgi-network, was examined in COS-7 cells. EGFP-Rap1A V12 was indeed distributed to the cytoplasmic compartment, mainly in the perinuclear region, where the trans-Golgi-network 38 protein exists (Fig. 4A). As is the case for phospholipase C⑀ (33), the interaction between Rap1A⅐GTP and the RA domain of RA-GEF-1 may cause translocation of RA-GEF-1 to perinuclear compartments. To test this possibility, EGFP-RA-GEF-1 or EGFP-RA-GEF-1⌬RA was coexpressed with HA-Rap1A V12 in COS-7 cells, and their localization was examined under a confocal microscope (Fig. 4B). EGFP-RA-GEF-1 and EGFP-RA-GEF-1⌬RA were evenly distributed in the cytoplasm without HA-Rap1A V12 (Fig. 4B, a and  b). When coexpressed with HA-Rap1A V12 , EGFP-RA-GEF-1 was observed in the perinuclear region, where HA-Rap1A V12 exists (Fig. 4B, c-e). In contrast, EGFP-RA-GEF-1⌬RA coexpressed with HA-Rap1A V12 remained in the cytoplasm and therefore did not colocalize with HA-Rap1A V12 (Fig. 4B, f-h). Similar results were obtained by using Rat-1 cells, in which expression levels of the exogenous proteins were lower than those in COS-7 cells (Fig. 4C).
Full-length RA-GEF-1 Associates with Neither cAMP nor cGMP-Previously, we could not detect the binding of cAMP or cGMP to the isolated cNMP-binding domain of RA-GEF-1 (24).
In this study, we tested whether full-length RA-GEF-1 is able to associate with either cAMP or cGMP. FLAG-RA-GEF-1 purified from Sf9 cells was immobilized on anti-FLAG M2 resin and examined for the association with [ 3 H]cAMP and [ 3 H]cGMP. The binding assay was performed at a lower salt concentration than that in the previous study (24) as described under "Experimental Procedures." Full-length RA-GEF-1 also exhibited no detectable binding to cAMP or cGMP (Fig. 5). Furthermore, no binding activity was detected under high salt conditions as described previously (24) (data not shown).

DISCUSSION
Diverse roles for Rap1 in cell signaling imply intricate mechanisms whereby Rap1 activity is strictly regulated by upstream signals. Three distinct signaling pathways mediated by cAMP, calcium, and diacylglycerol, respectively, have been implicated in Rap1 activation as shown by studies using second messenger analogues and specific inhibitors (2). Indeed, various types of GEFs for Rap1, which may play a role for Rap1 regulation in individual signaling pathways, have recently been identified. Epac/cAMP-GEF contains a cAMP-binding site through which cAMP induces GEF activity toward Rap1 (21,22). On the other hand, calcium-and diacylglycerol-binding domains were found in another Rap1 GEF, CalDAGGEFI (23). C3G also belongs to the Rap1 GEF family regulating Rap1 in response to tyrosine kinase-type receptor-mediated signals through interaction with the adaptor protein Crk (20).
A novel type of Rap1 GEF, RA-GEF-1, was cloned recently by our group and others (24 -27). RA-GEF-1 consists of multiple domains, including the cNMP-binding, Ras exchanger motif, PDZ, RA, and GEF domains, whose functions remain to be clarified. The cNMP-binding domain of RA-GEF-1 exhibits sequence homology to those of Epac/cAMP-GEF and the regulatory subunit of protein kinase A, except that the PRAA motif is missing. We and other groups observed no significant binding of cAMP and cGMP (24 -26), whereas one group described cAMP binding to this domain and a cAMP-dependent increase in the Ras⅐GTP level within CNrasGEF (RA-GEF-1)-expressing cells (27). However, the binding efficiency seemed very low, although the stoichiometry of cAMP binding was not clearly shown (27). Hence, it seems unlikely that cAMP directly regulates RA-GEF-1 in an allosteric manner. In this study, we further confirmed that full-length recombinant RA-GEF-1 as well as the isolated cNMP-binding domain does not show any significant cAMP or cGMP binding activity. RA-GEF-1 is unique in that it contains both the Rap1 GEF domain and the RA domain that binds Rap1⅐GTP, suggesting that it plays a role both upstream and downstream of Rap1. A possible mechanism for the regulation of GEF activity upon binding of Rap1 to the RA domain is a ligand-induced conformational change as proposed for Epac/cAMP-GEF (21,22). However, binding of Rap1 to the RA domain did not affect GEF activity in vitro, and therefore an allosteric effect may not be plausible. In contrast, as shown in this study, the RA domain contributes to direct the subcellular localization of RA-GEF-1, leading to the augmented activation of Rap1 at specific subcellular locations, presumably the Golgi apparatus and cytoplasmic vesicles. In other systems as well, the subcellular localization of an effector is determined through the interaction with a Ras family GTPase. For instance, the serine/threonine kinase Raf-1 translocates to the plasma membrane upon binding to Ras (34). Once targeted to the plasma membrane, Raf-1 becomes phosphorylated at multiple residues and activated (34). For the activation of Raf-1, however, conformational alteration caused by Ras⅐GTP binding is also considered to be crucial in addition to membrane recruitment (34,35). Phospholipase C⑀, a recently identified RA domain-containing phospholipase, is also activated through binding to Ras and Rap1 and subsequent translocation to the plasma membrane and the perinuclear region, respectively (33). Thus, a dual specificity RA domain like that of phospholipase C⑀ may play a pivotal role in switching and coordinating two signaling pathways mediated by distinct small GTPases by differentially distributing the effector. In addition, a molecule closely related to RA-GEF-1, named GFR/MR-GEF, was recently identified as a Rap1 GEF (36,37). The overall structural features of GFR/MR-GEF are similar to those of RA-GEF-1, and the RA domain specifically interacts with M-Ras⅐GTP. Although the mechanism remains to be clarified, the MR-GEFdependent accumulation of Rap1⅐GTP within the cell was abrogated by coexpression of the activated form of M-Ras (37).
In addition to the RA domain, the PDZ domain may serve to define the subcellular localization of RA-GEF-1. The PDZ domain of PSD95, for example, binds to a specific motif at the C terminus of several signaling molecules, such as the N-methyl-D-aspartate receptor and K ϩ channels (38). Through this interaction, a multimolecular signaling complex is composed at a specific site in the cell (38). Thus, it is conceivable that the subcellular localization of RA-GEF-1 is also regulated, at least in part, by the PDZ domain, although the binding partner is currently unidentified. Additional studies will reveal the precise mechanisms whereby the subcellular localization of Ras family GTPases and their signaling components is determined. Additionally, roles of Ras family GTPases specific to their subcellular localization will be clarified.