RalGEF2, a Pleckstrin Homology Domain Containing Guanine Nucleotide Exchange Factor for Ral*

Ral is a ubiquitously expressed Ras-like small GTPase. Several guanine nucleotide exchange factors for Ral have been identified, including members of the RalGDS family, which exhibit a Ras binding domain and are regulated by binding to RasGTP. Here we describe a novel type of RalGEF, RalGEF2. This guanine nucleotide exchange factor has a characteristic Cdc25-like catalytic domain at the N terminus and a pleckstrin homology (PH) domain at the C terminus. RalGEF2 is able to activate Ral both in vivo and in vitro. Deletion of the PH domain results in an increased cytoplasmic localization of the protein and a corresponding reduction in activity in vivo, suggesting that the PH domain functions as a membrane anchor necessary for optimal activity in vivo.

Ral is a ubiquitously expressed Ras-like small GTPase. Several guanine nucleotide exchange factors for Ral have been identified, including members of the RalGDS family, which exhibit a Ras binding domain and are regulated by binding to RasGTP. Here we describe a novel type of RalGEF, RalGEF2. This guanine nucleotide exchange factor has a characteristic Cdc25-like catalytic domain at the N terminus and a pleckstrin homology (PH) domain at the C terminus. RalGEF2 is able to activate Ral both in vivo and in vitro. Deletion of the PH domain results in an increased cytoplasmic localization of the protein and a corresponding reduction in activity in vivo, suggesting that the PH domain functions as a membrane anchor necessary for optimal activity in vivo.
Ral (RalA and RalB) is a small GTPase of the Ras family implicated in the control of cell proliferation (1-7), differentiation (8 -10), cytoskeletal organization (11), and vesicular transport (12). Ral is activated by a large variety of extracellular stimuli (13)(14)(15), and one of the mechanisms of Ral activation is the direct binding of active Ras to Ral-specific GEFs. 1 These GEFs, RalGDS, Rgl and Rlf, have a C-terminal Ras binding domain (RBD) responsible for the interaction with RasGTP and an N-terminal Cdc25-like catalytic domain (5,16,17). However, other mechanisms of Ral activation occur as well. In platelets and in fibroblasts, elevation of intracellular calcium levels induce Ral activation independently of Ras activation (13,18) and in neutrophils phosphatidylinositol 3-kinase (PI3K) and perhaps Src may mediate Ral activation (15). This suggests that, in addition to RalGDS, Rgl, and Rlf, other Ral regulatory proteins exist. Indeed, Rsc, a fusion protein isolated from a rabbit squamous cell carcinoma, has RalGEF activity and appears to lack an RBD (19). RalGTP levels are also regulated by GTPase-activating proteins (GAPs), including a high molecular mass RalGAP purified from brain and testis and a 43-kDa RalGAP in human platelets (20,21). Finally, RalA interacts with calmodulin (22,23), which may be involved in calcium-induced regulation of Ral (13,18).
In addition to proteins that regulate its activity, Ral interacts with several other proteins which may function as effectors. The active, GTP-bound form of Ral associates with RalBP (24 -26). RalBP is a GAP for Cdc42, a small GTPase involved in actin cytoskeleton organization and filopodia formation in fibroblasts (27). In addition, RalBP associates with Reps1 and POB1, two proteins implicated in endocytosis (12,28,29). Ral-GTP also binds to ABP280/filamin1, a cross-linker of actin filaments and a scaffold for several other proteins. This complex may mediate Cdc42-induced filopodia formation (11). Finally, Ral associates with the small GTPase Arf and phospholipase D in a nucleotide-independent manner (30 -33). Both phospholipase D and Arf have been implicated in many processes, including vesicular transport (34 -37). All these interactions point to a function of Ral in the control of the actin cytoskeleton and processes related to this, such as establishing cell polarity, migration, and vesicular transport (see Ref. 38).
In this paper we describe the identification of a novel Ralspecific GEF, RalGEF2. This GEF has a characteristic Cdc25like catalytic domain at the N terminus and a pleckstrin homology (PH) domain at the C terminus. This PH domain functions as a membrane anchor necessary for optimal activity of RalGEF2 in vivo.
Homology Searches-RalGEF2 was found by searching data bases for proteins containing Cdc25 catalytic domain homology. RalGEF2 was analyzed for protein domains using the ISREC ProfileScan server. The structurally conserved region (scr) sequence was defined from a comparison of various Ras and RalGEFs using the clustalX (40) alignment program, as was done for the alignment of homologous amino acids in the PH domain.
Cell Culture, Cell Lines, and Transfection-A14 (41) and Cos7 cells were grown at 37°C in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% heat-inactivated (30 min at 56°C) fetal calf serum and 0.05% glutamine. For activation studies, cells were serum-starved overnight for at least 16 h. Cells were transfected by the calcium-phosphate method.
Purification of GST Proteins-For purification of GST fusion constructs, protein expression was induced in DH5␣ using 100 nM isopropyl-1-thio-␤-D-galactopyranoside for 20 h at room temperature. Bacteria were collected and lysed in ice-cold phosphate-buffered saline containing 1% Triton X-100 and protease inhibitors. The lysates were sonicated three times for 20 s and centrifuged at 10,000 ϫ g for 20 min to remove insoluble material. GST fusion proteins were purified from the cleared lysate by batchwise incubation with glutathione-agarose beads (Sigma), and after washing the protein was eluted from the beads in buffer containing 50 mM Tris, pH 7.5, 100 mM NaCl, 10% glycerol, and 10 mM glutathione. The eluted protein was dialyzed for 20 h in the same buffer without glutathione. The purification procedure of the small GTPases used in the in vitro experiments is described elsewhere (42)(43)(44).
Immunoprecipitation and Western Blotting-Western blotting of all protein samples was performed using polyvinylidene difluoride membranes. The antiserum used for detecting endogenous RalGEF2 was raised against a synthetic peptide that consists of the amino acids 543-557 (KSNRPQVPANLMSFE) of RalGEF2. Other antibodies used are anti-HA (12CA5) (45), anti-Ral, anti-Ras, and anti-Rap (all three from Transduction Laboratories). Anti-RalGEF2 and 12CA5 were also used to perform immunoprecipitations.
In Vivo Activation of Small GTPases-Cells were transiently transfected with HA-tagged versions of the small GTPases either alone or in combination with RalGEF2, and serum-starved for 20 h prior to lysis. GTP-bound forms of the different GTPases were isolated using activation-specific probes and subsequently quantified, as described (13,46,47). For RalGTP the Ral-binding domain of RLIP was used, for Ras the RBD of Raf1, and for Rap the RBD of RalGDS.
In Vitro Activation of Small GTPases-In vitro GEF activity was measured as described (48). Briefly, 250 nM purified GTPase, loaded with fluorescently labeled 2Ј,3Ј-bis(O)-N-methylantharanoloyl guanosine diphosphate (mantGDP), was incubated, in the presence of excess unlabeled GDP (5 mM), with 50 nM of purified GST-cat in 50 mM Tris, pH 7.5, 5 mM Mg 2ϩ , and 5 mM dithioerythritol at 20°C. Release of mantGDP was measured in real time as a decrease in fluorescence using a Perkin-Elmer fluorometer LS50B (excitation wavelength 366 nm, emission wavelength 450 nm). This decrease in fluorescence is caused by quenching of released mantGDP by water. In the Ral assay, 0.15 mg/ml bovine serum albumin was added to stabilize the protein.
Alternatively, we used a reverse assay (49) in which 150 nM RalGDP was added to a cuvette containing 100 nM mantGDP either alone or in Subcellular Fractionation-Cells were harvested in lysis buffer (20 mM Hepes, pH 7.4, 5 mM EGTA, 1 mM sodium vanadate, 1 M leupeptin, 0.1 M aprotinin) and subsequently homogenized through a 23-gauge 1.25 Microlance syringe. Intact cells and nuclear components were removed by a sequential two-step centrifugation at 6000 rpm for 1 min (Eppendorf table centrifuge). Subsequently, the samples were centrifuged at 100,000 ϫ g at 4°C for 90 min. The supernatant was collected as the soluble fraction, and the particulate fraction was dissolved in buffer containing 1% Triton X-100, 50 mM Hepes, pH 7.5, 50 mM NaCl, 5 mM EDTA, 1 mM vanadate, 1 M leupeptin, and 0.1 M aprotinin. Soluble and particulate fractions were analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotting.

RESULTS
A Novel Ral Guanine Nucleotide Exchange Factor-In our ongoing search for regulators of Ras family members, we found a protein in the data base from the Kazusa DNA Research Institute (KIAA0351) with similarities to RasGEFs. A schematic representation of the non-coding and coding sequences of KIAA0351 is presented in Fig. 1A. As shown below, this protein is a GEF for the small GTPase Ral; hence, we named it Ral-GEF2, since this GEF forms a novel subclass, distinct from the RalGEF members, which are characterized by the presence of the RBD domain (Fig. 1B). RalGEF2 has the characteristic GEF domain present in all RasGEFs and shows considerable homology to previously described RasGEFs, especially in the structurally conserved regions scr1, scr2, and scr3 (data not shown). In addition, RalGEF2 has a PH domain in its C terminus that shows highest homology to the PH domain in the Drosophila RhoGEF Still Life (Sif) (50), and the N-terminal PH domains in the RacGEFs Tiam1 and Stef (51, 52) (Fig. 1C). Surprisingly, no Ras exchange motif (REM) is present in Ral-GEF2, in contrast to all other Cdc25-like GEFs identified so far.
Expression of RalGEF2-A polyclonal antibody (␣-RalGEF2) was raised against a C-terminal peptide of RalGEF2. This antibody recognized the 60-kDa HA-RalGEF2 protein immunoprecipitated from Cos7 cell lysate transiently transfected with HA-RalGEF2 using an ␣-HA-monoclonal antibody (Fig. 2). A similar-sized protein was identified in an ␣-RalGEF2 immunoprecipitate from 293 cells. Both protein bands disappeared when the antibody was preincubated with the immunizing peptide. From this result, we concluded that ␣-RalGEF2 recognizes RalGEF2 both in immunoprecipitation and blotting experiments. Next, a Western blot containing protein samples from various human tissues was probed with either ␣-RalGEF2 or ␣-RalGEF2 blocked with peptide. As shown in Fig. 2B, RalGEF2 expression is rather ubiquitous. It is high in brain, heart, kidney, adrenal gland, and colon; low in pancreas, skeletal muscle, thymus, and liver; and intermediate in lung and spleen. Ubiquitous expression of RalGEF2 was also shown using reverse transcription-polymerase chain reaction by the Kazusa DNA Research Institute.
RalGEF2 Activates Ral Both in Vivo and in Vitro-To investigate which Ras-like GTPase is activated by RalGEF2, we incubated GST-cat (Fig. 3A) with various Ras-like GTPases loaded with fluorescent mantGDP. Release of guanine nucleotide bound to Ral was measured in real time as decrease in fluorescence. As shown in Fig. 3B, incubation of Ral-mantGDP alone already resulted in a decrease in fluorescence, indicating that a fraction of the protein sample was unstable and degraded during the incubation. In the presence of GDP, intrinsic exchange activity is measured. We observed a clear increase in exchange when GST-cat was added as well, showing that GSTcat catalyzes Ral guanine nucleotide exchange. GST-cat did not affect the exchange rate of either Ras or Rap2 (Fig. 3C). As an alternative assay, we used a reverse procedure in which binding of mantGDP to Ral results in an increase in fluorescence. mantGDP was incubated either alone, in the presence of GSTcat or in the presence of GST-Rlf and RalGDP was added. As shown in Fig. 3D, both GST-cat and GST-Rlf catalyzed guanine nucleotide exchange. Surprisingly, when a shorter version of Ral lacking an additional 28 residues at the C terminus was used in the assay, GST-cat failed to induce exchange, in contrast to GST-Rlf. This effect was also observed when corresponding Drosophila Ral proteins were used (data not shown). To investigate whether GST-cat may still bind to Ral, we preincubated Ral-mantGDP and GST-cat for prolonged period of time followed by the addition of GST-Rlf. GST-cat did not affect the ability of GST-Rlf to induce exchange (Fig. 3E), indicating that RalGEF2 fails to interact stably with the C-terminally truncated protein. From these results we conclude that RalGEF2 is indeed a GEF for Ral, and that, in contrast to Rlf, RalGEF2 requires the C-terminal 28 residues of Ral for proper interaction.
Subsequently, we analyzed whether RalGEF2 also activates Ral in vivo. HA-RalGEF2 was cotransfected with HA-tagged RalA in Cos7 cells, and Ral activity was measured using activation-specific probes. As shown in Fig. 4A, cotransfection with RalGEF2 resulted in a clear activation of Ral. In contrast, RalGEF2 displayed only marginal activity for HA-Ras and no exchange activity for HA-Rap1 in vivo (Fig. 4, B and C). From this result we conclude that RalGEF2 selectively activates Ral in vivo. Next, we addressed the question whether indeed the GEF activity is responsible for this effect. HA-cat-CAAX, which is described below, activates cotransfected HA-Ral. We made mutants of this protein in a region that is fully conserved in all human RalGEFs, located in the scr2, in order to abolish its catalytic activity. We found that both HA-cat L148A and HA-cat ⌬SALQS (residues 146 -150) failed to activate HA-Ral  (Fig. 4D). From these data we conclude that activation of Ral in vivo is directly mediated by the catalytic activity of RalGEF2.
Regulatory Function of the PH Domain-PH domains commonly interact with membrane lipids, in particular phosphorylated phosphatidylinositol lipids. As such, PH domains function either as membrane anchor or as a regulatory domain that responds to the products of PI3K, phosphatidylinositol 3,4bisphosphate, or phosphatidylinositol 3,4,5-trisphosphate. To determine whether RalGEF2 responds to PI3K signaling, we introduced RalGEF2 in A14 cells and stimulated the cells with insulin, a potent inducer of PI3K activity (45). However, we did not observe any increase in RalGEF2 activity by insulin treatment (data not shown). Additionally, other stimuli tested, including epidermal growth factor, endothelin, forskolin, ionomycin, lysophosphatidic acid, and serum, failed to activate RalGEF2. We therefore investigated whether the PH domain is involved in membrane localization of RalGEF2. Cells were transfected with HA-RalGEF2 or mutants lacking the PH domain (HA-⌬PH and HA-cat), and the presence of these proteins in the cytosol and the membrane-enriched particulate fraction was determined. Deletion of the PH domain resulted in a clear reduction of the level of RalGEF2 in the membrane fraction (Fig. 5A), indicating that the PH domain is involved in membrane localization of RalGEF2. This reduced association to membranes could be restored by adding the C-terminal polybasic domain and CAAX sequence of Ki-Ras to HA-⌬PH and HA-cat (HA-⌬PH-CAAX and HA-cat-CAAX). This region directs the addition of a C-terminal isoprenyl group and as a consequence membrane attachment (Fig. 5A).
Next, we investigated whether deletion of the PH domain also affects GEF activity in vivo. Cells were cotransfected with the various RalGEF2 constructs and HA-Ral, and the level of RalGTP was determined. Even at lower levels of expression (Fig. 5B, upper panel), full-length RalGEF2 is much more efficient in activating Ral than the mutant lacking the PH domain (Fig. 5B), showing that the PH domain is required for efficient RalGEF2 activity. The reduced GEF activity of HA-⌬PH can be restored by the addition of a membrane anchor (HA-⌬PH-CAAX; Fig. 5C). Similarly, addition of the CAAX domain to HA-cat increased the efficiency of Ral guanine nucleotide exchange activity (Fig. 5D). From these results we conclude that the PH domain of RalGEF2 is a membrane-targeting sequence responsible for efficient RalGEF activity. DISCUSSION We have identified a novel GEF for the small GTPase Ral, RalGEF2. This is a ubiquitously expressed protein that is particularly abundant in brain, heart, kidney, adrenal gland, and colon. Previously, four other RalGEFs have been identified, including RalGDS, Rgl, and Rlf. These GEFs have in addition to the catalytic region a C-terminal Ras binding domain. The most striking characteristic of RalGEF2 is its PH domain. This   1, open circles), with 5 mM GDP to measure the domain is most similar to PH domains present in the Rho/ RacGEFs Tiam1, SIF, and Drosophila Still life. PH domains may bind to phosphatidylinositol lipids, which are either constitutively present or which are induced by certain stimuli. For instance, the product of PI3K, phosphatidylinositol 3,4bisphosphate, recruits target proteins to the membrane by binding to their PH domains (53,54) Via this relocalization, a protein is for instance either brought into the vicinity of its target or of its activators. However, the fact that insulin, a very strong inducer of PI3K, fails to activate RalGEF2 suggests that RalGEF2 activity is not induced by these lipids. Thus, the PH domain may serve to constitutively anchor RalGEF2 to mem-branes. Indeed, deletion of the PH domain results in an increased cytoplasmic localization of the protein and a corresponding reduction in the efficiency of Ral activation in vivo. Both membrane localization and efficient GEF activity could be restored by the addition of the C-terminal polybasic region and CAAX motif of Ki-Ras, which targets proteins to the membranes. From these results, we conclude that the PH domain is predominantly responsible for membrane localization of RalGEF2.
Our failure to induce the activity of RalGEF2 by external stimuli may indicate that RalGEF2 is a constitutively active GEF, or that we have not yet identified the proper stimulus. Whereas Ral clearly serves as a downstream target of Ras signaling, through the direct binding of members of RalGDS, Rgl and/or Rlf to activated Ras, it is clear from several studies that other pathways are also mediating Ral activation. For instance, calcium can activate Ral independently of Ras (13,18), while in neutrophils as yet unidentified pathways exist (15). RalGEF2 may mediate one of these pathways.
RalGEF2 is distinct from the other RalGEFs in that it does not have a REM domain. This REM domain is thought to play a role in stabilizing the catalytic domain of Ras-like GEFs. Perhaps the PH domain or another region serves a similar function. An additional striking difference between RalGEF2 and other RalGEFs, in particular Rlf, is that RalGEF2 does not exchange nucleotide in vitro from a truncated version of Ral. Apparently the C-terminal region of Ral contains amino acids essential for the proper binding of the catalytic domain of RalGEF2. Whether RalGEF2 directly interacts with this region, or if the absence of the REM domain is responsible for this difference awaits further investigation. However, this result clearly indicates that RalGEF2 uses a different molecular mechanism to release GDP from Ral than Rlf does.
Ral has been implicated in a variety of cellular processes. Most notable is the role of Ral in coupling signals from Ras to the induction of transcription, such as transcription from serum response elements (5,55) and the inhibition of the transcription factor AFX (56). These effects may, at least in part, explain the effects of Ral-mediated signaling on cell proliferation (1-7) and differentiation (8 -10). However, the mechanism by which Ral regulates transcription is unclear. Another function of Ral is the regulation of the cytoskeleton. This is indicated by the association of the active form of Ral with RalBP, a GAP for the small GTPase Cdc42, and to filamin/ABP280, a protein involved in the cross-linking of actin filaments. Indeed, Ral was found to mediate Cdc42-induce filopodia formation. Finally, Ral has been implicated in vesicular trafficking. Ral localizes in part in vesicular membranes, and, recently, it was shown that Ral, through RalBP and the RalBP-associated proteins Reps1 and Pob1, is involved in the regulation of receptor endocytosis. In addition, Ral mediates Fc⑀RI-induced histamine secretion (15). It may be that all these effects are pleiotropic, resulting from the activation of a single pool of active Ral. As a consequence, RalGEF2 may be involved in each of these processes. Alternatively, the distinct regulatory mechanism of RalGEF2 exchange activity, PH domain-dependent membrane localization, suggests the possibility that compartmentalized activation of Ral may regulate distinct functions of Ral.