Identification and characterization of a new family of guanine nucleotide exchange factors for the ras-related GTPase Ral.

Guanine nucleotide exchange factors (GEFs) are responsible for coupling cell surface receptors to Ras protein activation. Here we describe the characterization of a novel family of differentially expressed GEFs, identified by database sequence homology searching. These molecules share the core catalytic domain of other Ras family GEFs but lack the catalytic non-conserved (conserved non-catalytic/Ras exchange motif/structurally conserved region 0) domain that is believed to contribute to Sos1 integrity. In vitro binding and in vivo nucleotide exchange assays indicate that these GEFs specifically catalyze the GTP loading of the Ral GTPase when overexpressed in 293T cells. A central proline-rich motif associated with the Src homology (SH)2/SH3-containing adapter proteins Grb2 and Nck in vivo, whereas a pleckstrin homology (PH) domain was located at the GEF C terminus. We refer to these GEFs as RalGPS 1A, 1B, and 2 (Ral GEFs with PH domain and SH3 binding motif). The PH domain was required for in vivo GEF activity and could be functionally replaced by the Ki-Ras C terminus, suggesting a role in membrane targeting. In the absence of the PH domain RalGPS 1B cooperated with Grb2 to promote Ral activation, indicating that SH3 domain interaction also contributes to RalGPS regulation. In contrast to the Ral guanine nucleotide dissociation stimulator family of Ral GEFs, the RalGPS proteins do not possess a Ras-GTP-binding domain, suggesting that they are activated in a Ras-independent manner.

into the pFLAG-CMV2 vector (Eastman Kodak Co.), creating FLAG-RalGPS 1A. Full-length human RalGPS 1B (KIAA0351) was obtained from Kazusa DNA Research Institute, Chiba, Japan. The FLAG-tagged construct was prepared by purifying the XbaI-SmaI fragment of Ral-GPS 1B and inserting it into the XbaI-SmaI site of pFLAG-CMV2 RalGPS 1A from above. The C termini (residues 392-529 and 433-557 for RalGPS 1A and B, respectively) encompassing the PH domain were removed from RalGPS 1A/B by digesting with NcoI, blunting, and inserting the fragment (ClaI-NcoI/blunt) into the ClaI-EcoRV sites of pFLAG-CMV2. The CAAX construct was generated by inserting the above (ClaI-NcoI/blunt) fragment into the ClaI-EcoRV sites of pE-CAAX vector (created by G. Clark, NCI, National Institutes of Health). RalGPS 1A/B-⌬PH-CAAX was then directionally ligated into the BamHI site of pFLAG-CMV2.
Ras Protein Binding Assay-Three 100-mm dishes of 70% confluent 293T cells were transfected with 7.5 g of FLAG-RalGPS 1A/B using 15 g of NovaFECTOR/dish (Venn Nova Inc.) following the manufacturer's directions. After 72 h, cells were lysed (1 ml/100-mm dish) in 20 mM Tris-HCl, pH 7.4, 1% IGEPAL (Sigma), 1 mM EDTA, 100 mM NaCl, 0.05 TIU/ml of aprotinin, and 1 mM PMSF. Ras-bound beads containing ϳ20 g of protein were mixed with empty glutathione-agarose beads (50% slurry) to give a final volume of 35 l. The beads were rinsed twice with 1 ml of Ras buffer (20 mM Tris-HCl, pH 7.6, 5% glycerol, 50 mM NaCl, 0.1% Triton X-100, and 1 mM DTT) and then incubated with the Ras buffer spiked with 10 mM EDTA for 10 min at room temperature. Approximately 400 l of cell lysate containing FLAG-RalGPS 1A/B was added to the rinsed Ras-bound beads. This slurry was tumbled for 2 h at 4°C and then the beads were washed 4 times with Ras buffer supplemented with EDTA to 5 mM.
In Vivo Exchange Assay-293T cells were transfected with 750 ng of pFLAG-CMV2-RalA and 750 ng of the indicated GEF using 8 l of NovaFECTOR/60-mm dish. After 24 h, serum was removed for another 18 h. Cells were rinsed twice in phosphate buffered saline, lysed with Ral buffer (0.5 ml/dish; 50 mM Tris-HCl, pH 7.4, 10% glycerol, 200 mM NaCl, 2.5 mM MgCl 2 , 1% IGEPAL (Sigma), aprotinin (0.05 TIU/ml), 1 mM PMSF), and cleared by centrifugation for 10 min at 12,000 rpm. Approximately 10 g of GST-RID or -Raf (residues 2-140) bound to 35 l of glutathione-agarose beads (50% slurry) were rinsed twice with Ral buffer and then tumbled with 400 l of cell lysate for 1 h at 4°C. The beads were washed 4 times with Ral buffer, omitting protease inhibitors. 50 l of SDS-PAGE sample buffer was added to the beads, and the activated Ral proteins bound to GST-RID were visualized by immunoblotting of the FLAG-tagged Ral proteins.
Coimmunoprecipitation-293T cells were cotransfected with 2 g of pFLAG-CMV2-RalGPS 1B and 2 g of either pCGN-Grb2 or -Nck using 15 l of NovaFECTOR/100-mm dish. After 72 h the cells were lysed in 1 ml of Ral buffer. Cellular debris was removed by a 3 min, 13,000 ϫ g spin, and the cellular lysate was precleared for 15 min at 4°C with 30 l of protein A/G-Sepharose (Santa Cruz Biotechnology, Inc.). The precleared lysate was tumbled with 3 l of M2 anti-FLAG antibody (Sigma) for 5 h at 4°C. The immunocomplex was precipitated by a 30-min incubation with 50 l of protein A/G-Sepharose. The Grb2 and Nck proteins that coprecipitated with Flag-RalGPS 1B were visualized by immunoblotting with anti-HA antibody (Babco) following SDS-PAGE.
Levels of RalGPS 1B expression were determined by M2 anti-FLAG antibody using ECL reagents (Amersham Pharmacia Biotech).

RalGPS Represents a Novel Family of Putative Ras GEFs-
The catalytic (CDC25 homology) domains of Ras family GEFs share ϳ30% homology with each other. Highest sequence homology is found within SCRs (structurally conserved regions) 1-3 as defined in Ref. 8. Using a consensus SCR2 sequence supplemented with GRF1 residues at positions of low homology, we searched the EST database and identified several novel putative GEF sequences. One of these, AA121710, was obtained, sequenced, and found to encode the C-terminal portion of a CDC25 homology domain, as well as a proline-rich region similar to SH3 binding motifs, a pleckstrin homology domain, a stop codon, and a 3Ј non-coding region (Fig. 1). This protein is referred to as RalGPS 1A. Using 5Ј RACE we obtained the core catalytic domain (SCRs 1-3) but could not isolate SCR0 (also known as REM, Ras exchange motif or the conserved noncatalytic domain (15)), an additional N-terminal domain found in most GEFs and shown in the Sos1 crystal structure to play a role in maintaining the conformation of the catalytic domain (16). Subsequently an additional full-length sequence, KIAA0351 (accession number AB002349), which encoded a splice variant of RalGPS 1A, referred to subsequently here as RalGPS 1B, was entered in the NCBI database. The 5Ј end of this clone extended beyond our 1A sequence and revealed a stop codon 5Ј to a potential initiating ATG. This indicated that like the yeast GEF, BUD5, the RalGPS proteins do not contain the SCR0 domain (15). RalGPS 1B contained an extra 8-amino acid insertion within loop 3 of its PH domain, a shorter, alternate C terminus, and a unique 69-amino acid central portion that replaced 27 amino acids in RalGPS 1A (Fig. 1A). An additional mouse sequence, AA110466, appears to represent a novel family member, RalGPS 2 ( Fig. 1B), because a closely related partial human sequence (AA252781) is also present in the database. The AA110466 partial cDNA was obtained and sequenced and found to similarly contain a PH domain (Fig.  1C) and PXXP motif. SCRs 1-3 were originally defined primarily using yeast Ras GEF sequences (8). Alignment with subsequently identified mammalian GEF sequences has revealed two other regions of high homology that are indicated as SCRs 4 and 5 in Fig. 1B. Interestingly, SCR4, a highly conserved I/VNF motif has been reported to sit within a small hydrophobic groove formed by the SCR0 domain in the Sos1 crystal structure (16). Because SCR0 is absent from RalGPS, it is not surprising that charged residues (ENE) occupy the place of the INF motif of Sos1 (Fig. 1B). Further, RalGPS family members do not have the extension in SCR3 characteristic of the RalGDS family of Ral GEFs (Fig. 1B).
Splice Variants of RalGPS 1 Are Differentially Expressed-A 1.5-kb XbaI-XhoI fragment of AA121710 encompassing most of the RalGPS 1A/B coding region was used to probe a human multitissue Northern blot to examine mRNA expression. Three bands were observed (Fig. 2), presumably because of alternate splicing of the message. Because the full-length KIAA0351 cDNA is 6336 base pairs in length, RalGPS 1B likely represents the middle, ϳ7-kb band that is ubiquitously expressed. Assuming that RalGPS 1A and B have identical 5Ј ends the 1A message is predicted to be ϳ2 kb and is therefore represented by the lower, 2.6-kb blot band that was preferentially expressed in heart and testis. The identity of the larger ϳ12-kb doublet was not established. Therefore, similarly to GRF1 (17), there are multiple spliced variants of the RalGPS GEFs that are differentially expressed in human tissues.
RalGPS 1 Is a Ral-specific GEF-To identify Ras substrates for RalGPS 1A and B, we first assessed their ability to bind a panel of nucleotide-free GTPases. Because GEFs bind with high affinity to Ras proteins in the absence of Mg 2ϩ /GTP or GDP and stabilize the nucleotide-free state (2), epitope-tagged RalGPS 1 proteins expressed in 293T cells were incubated with GST-Ras fusion proteins that had been stripped of GDP in the presence of EDTA. As shown in Fig. 3A, nucleotide-free Ral but not other GTPases or GST bound to both RalGPS 1 splice variants.
To confirm that the interaction of RalGPS with Ral resulted in nucleotide exchange we examined the ability of RalA to be loaded with GTP in vivo following cotransfection with known and putative Ral GEFs. For these assays we took advantage of the fact that the putative Ral effector RalBP1 binds preferentially to Ral-GTP versus Ral-GDP to specifically extract activated Ral from cell lysates (18,19). As can be seen in Fig. 3B, there was little Ral-GTP present in serum-starved 293T cells. However, following co-transfection with the RalGDS family member Rlf or with RalGPS 1B, Ral-GTP levels were increased by an equivalent amount, indicating that RalGPS was indeed a Ral GEF. Because a trace of Rap1A binding to RalGPS was detected in some assays (see Fig. 3A), we examined the Ras protein nucleotide exchange specificity of RalGPS. Using a GST fusion protein containing the Ras/Rap1-GTP-binding domain of cellular Raf1 to precipitate activated Rap1A from cell lysates we found that cotransfection with RalGPS 1B failed to activate Rap1A under conditions where the known Rap1 GEF, C3G, promoted significant accumulation of Rap1A-GTP in vivo. Taken together, the data in Fig. 3, A and B indicate that Numbers indicate the extent of amino acid divergence between the 1A (left) and B (right) splice variants. B, sequence alignment of GEF domains from human RalGPS 1A/B (AB002349) and a partial mouse RalGPS 2 clone (AA110466) with Ral GEFs (mouse RalGDS, L07924; mouse Rgl, U14103; mouse Rlf, U54639; and rabbit Rsc, U82163), Ras GEFs (human Sos1, L13857 and rat GRF1, X67241), and the Rap1 GEF, C3G (human, D21239). Sequences were aligned using Clustal W with subsequent manual adjustment. Similarities were highlighted using MacBoxshade. C, sequence alignment of RalGPS PH domains with the N-terminal PH domains of Tiam-1 (Q60610), Still life 1 (P91621), pleckstrin (P08567), and GRF1 (X67241). A potential phosphorylation site in RalGPS 1B is underlined.

FIG. 2. Multitissue Northern blot showing RalGPS 1A/B distri-
bution. An ϳ1.5-kb XbaI-XhoI fragment encompassing most of the RalGPS 1A coding region was used to screen a human multitissue Northern blot essentially as described by the manufacturer. Two bands were predicted to be RalGPS 1A and B based on the sizes of their respective cDNAs. skel. muscle, skeletal muscle; sm. intestine, small intestine; m.l., mucosal lining; p.b., peripheral blood. RalGPS 1 is a highly specific Ral GEF. We have not confirmed that RalGPS 2 is also a Ral GEF, but given the high degree of sequence conservation with RalGPS 1A/B within its catalytic domain (Fig. 1B) this seems likely.
The PH Domain of RalGPS Is Required for Biological Activity-PH domains are typically involved in protein-protein or protein-lipid interactions (10). For example, the PH domains of the ␤-adrenergic receptor kinase and GRF bind to the ␤␥ subunits of heterotrimeric G proteins (20), whereas most PH domains associate with the polyphosphoinositides, phosphatidylinositol 4,5 bisphosphate and phosphatidylinositol 1,4,5trisphosphate (PIP 3 ) (21). Comparison of the RalGPS PH domain with that of other molecules demonstrated strongest identity with the N-terminal PH domains of Tiam-1 and the Drosophila melanogaster Still life (Fig. 1C), GEFs that regulate Rho family GTPases (22,23). It has been reported that the Tiam-1 N-terminal PH domain binds to PIP 3 (21) and is required for membrane association and biological activity of this Rac GEF (23). Therefore, we deleted the PH domain from RalGPS 1B (RalGPS-⌬PH) and examined the consequence of this modification on in vivo GEF activity.
As shown in Fig. 3C, this deletion completely abrogated the ability of overexpressed RalGPS 1B to activate Ral in vivo. To address whether the PH domain might be responsible for interaction of the GEF with the plasma membrane, it was replaced with the C-terminal 18 residues of Ki-Ras that we and others have previously reported will target heterologous proteins to the plasma membrane (24,25). Upon addition of this Ras CAAX sequence to RalGPS-⌬PH, we rescued its ability to promote Ral-GTP accumulation in vivo (Fig. 3C), suggesting that the role of the PH domain was indeed membrane targeting. It has previously been established that Ras is responsible for the recruitment of RalGDS family members to the plasma membrane and that this membrane targeting is essential for GEF activity (26,27). The RA domain responsible for RalGDS association with Ras-GTP (11, 12) is absent from the RalGPS family and is presumably replaced in RalGPS by the PH domain. We have not established whether the sequence difference between RalGPS 1A and B in loop 3 of the PH domain affects their ligand specificity or function. However, the sequence of RalGPS 1B includes a consensus phosphorylation site for the cyclic AMP-dependent protein kinase (KKVSI (28), underlined in Fig. 1C). Introduction of a negatively charged phosphate group at this site could inhibit binding of the PH domain to PIP 3 . Although full-length RalGPS 1A was expressed poorly in 293T cells, RalGPS 1A-⌬PH-CAAX activated Ral as effectively as its 1B counterpart. 2 The RalGPS 1 Proline-rich Motif Binds to Grb2 and Nck and Facilitates Adapter Protein-mediated Activation of Ral-As shown in Fig. 4A, the proline-rich motif present in RalGPS 1A/B bears sequence homology with the SH3 binding motifs of the Grb2 and Nck adapter protein targets, Sos1 and PAK. In addition to the core PXXP motif, there are surrounding proline residues to form a left-handed P3 helix plus adjacent arginine residues that help orient the peptide and confer binding specificity (9). Indeed the proline-rich region in RalGPS matches consensus sequences derived for Grb2 and Nck SH3 binding peptides by phage-displayed random peptide libraries (9,14). To examine the ability of RalGPS 1 to bind to SH3 domaincontaining proteins in vitro, FLAG-tagged RalGPS 1B was FIG. 3. RalGPS 1 is a Ral GEF that requires membrane targeting to function in vivo. A, the indicated GST-Ras fusion proteins (20 g) were stripped of nucleotide and used to precipitate FLAG-tagged RalGPS 1A or B from 293T cell lysates. Bound GEFs were detected by Western blotting precipitates with M2 anti-FLAG antibody. B, empty vector or vectors encoding RalGPS 1B, the Ral GEF, Rlf, or the Rap1 GEF, C3G were cotransfected into 293T cells with vectors encoding FLAG-tagged RalA or Rap1A as indicated. Following overnight serum starvation, cells were lysed, and RalA-GTP and Rap1A-GTP were extracted by incubation with the GTPase-binding domains of RalBP1 and cellular Raf1, respectively. Stimulation of Ral/Rap1-GTP accumulation was determined by Western blotting with anti-FLAG antibody. Lower panels demonstrate equal GTPase transfection levels under each condition by Western blotting cell lysates for FLAG Ral or Rap1. C, 293T cells were transfected with vectors encoding FLAG-tagged RalA Ϯ the indicated FLAG-tagged Ral GEFs. Deletion of the PH domain (RalGPS-⌬PH) resulted in loss of in vivo exchange activity, whereas attachment of the Ki-Ras C terminus (RalGPS-⌬PH-CAAX) resulted in its restoration. All data are representative of at least three independent experiments.

FIG. 4. The RalGPS 1 PXXP motif is required for Grb2 and Nck SH3 domain binding and regulation.
A, alignment of proline-rich motifs that bind to the N-terminal Grb2 or central Nck SH3 domains. B, the ability of FLAG-RalGPS 1B to bind to SH3 domains in vitro was determined by incubating 293T cell lysates with equal amounts of glutathione-agarose bead-immobilized GST fusion proteins containing full-length Grb2, the SH3 domains of Nck, the isolated SH3 domains from p120 Ras GAP, Abl, Src, and phospholipase C␥, or the N-terminal SH3 domains from Crk and p67 phox . Beads were washed, and associated RalGPS was detected by Western blotting. C, to determine whether RalGPS could interact with SH3-containing adapter proteins in vivo, 293T cells were cotransfected with empty vector or vector encoding FLAG-RalGPS 1B, HA-Grb2, or HA-Nck as indicated. Following immunoprecipitation with anti-FLAG antibody, associated adapter proteins were detected by blotting with anti-HA antibody. Lower panels demonstrate expression levels of RalGPS, Grb2, and Nck in cell lysates. HC and LC indicate heavy and light chains, respectively, of the M2 anti-FLAG antibody. D, RalGPS-⌬PH is activated by Grb2. 293T cells were cotransfected with plasmids encoding FLAG-RalA, RalGPS-⌬PH, and Grb2 as indicated. After serum starving over night cell lysates were prepared, and RalA-GTP was precipitated using GST-RalBP1(10 g) and detected by Western blotting. All data are representative of at least three independent experiments. expressed in 293T cells and cell lysates incubated with GST fusion proteins containing the SH3 domains of various proteins. As seen in Fig. 4B, RalGPS 1B was precipitated by SH3 domains from Grb2, Nck, Src, and PLC␥ but not those of p120 Ras GAP, Crk, Abl, or p67 phox . To determine whether the interactions with adapter proteins might occur in vivo, 293T cells were co-transfected with plasmids encoding FLAG-tagged RalGPS 1B and HA-tagged Grb2 or Nck. Following lysis and precipitation with anti-FLAG antibody, both Grb2 and Nck were found to coprecipitate with RalGPS (Fig. 4C).
It is well documented that Grb2 contributes to the recruitment of the Ras GEF, Sos1, to the plasma membrane, leading to Ras activation (2) and that the Sos PH domain also plays a role in this process (29). Association of Grb2 with Sos may also alleviate negative allosteric control imparted by the Sos C terminus (25,30). Although over-expression of Grb2 did not enhance the ability of recombinant RalGPS to activate Ral in vivo (Fig. 4), we rationalized that the adapter protein/GEF interaction likely contributes to the physiological regulation of RalGPS. Because RalGPS did not activate Ral following removal of the GEF's PH domain, we examined whether overexpression of Grb2 might enhance the activity of this ⌬PH mutant. As seen in Fig. 4D, co-transfection with Grb2 did indeed promote Ral activation by RalGPS-⌬PH, indicating that, as described for Sos1, growth factor stimulation may be responsible for promoting the translocation and/or activation of RalGPS. Because Grb2 alone did not promote Ral-GTP accumulation we can conclude that this effect was dependent on the presence of transfected RalGPS and was not due to activation of an endogenous Grb2ϾSosϾRasϾRalGDSϾRal pathway. This notion was further supported by failure of a dominant inhibitory Ras(17A) mutant to block the ability of Grb2 to activate Ral in the presence of RalGPS-⌬PH. 2 Although Ras could potentially modulate full-length RalGPS activity in vivo via a PI3K-PIP 3 pathway, surprisingly, we did not see an effect of a PI3K inhibitor (LY294002, 30 M) on Ral activation by this GEF. As seen in Fig. 4A, RalGPS 2 deviates from 1A and B in residues surrounding the PXXP motif. Whether this deviation affects the ability of RalGPS 2 to differentially interact with SH3 domain-containing adapter proteins will require further investigation.
In conclusion, we have identified a novel family of putative GEFs that includes at least two members plus differentially expressed splice variants. RalGPS 1 specifically associated with and activated the Ras-related GTPase, Ral, when overexpressed in 293T cells. In contrast to the RalGDS family of Ral GEFs that rely on interaction with Ras-GTP for membrane translocation and Ral activation (27), but similarly to the Ras GEF, Sos1 (30), RalGPS 1 nucleotide exchange activity is regulated in vivo by a PH domain and a Grb2-binding PXXP motif. Although Ca 2ϩ has been reported to activate Ral in a Rasindependent manner (31) there is no indication from their primary sequences that RalGPS 1A/B could respond to Ca 2ϩ elevation. Why nature has created so many Ral GEFs and how they are differentially regulated will be the topics of future investigation.