Calmodulin Binds RalA and RalB and Is Required for the Thrombin-induced Activation of Ral in Human Platelets*

Ral GTPases may be involved in calcium/calmodulin-mediated intracellular signaling pathways. RalA and RalB are activated by calcium, and RalA binds calmodulin in vitro. It was examined whether RalA can bind calmodulin in vivo, whether RalB can bind calmodulin, and whether calmodulin is functionally involved in Ral activation. Yeast two-hybrid analyses demonstrated both Rals interact directly but differentially with calmodulin. Coimmunoprecipitation experiments determined that calmodulin and RalB form complexes in human platelets. In vitro pull-down experiments in platelets and in vitrobinding assays showed endogenous Ral and calmodulin interact in a calcium-dependent manner. Truncated Ral constructs determined in vitro and in vivo that RalA has an additional calmodulin binding domain to that previously described, that although RalB binds calmodulin, its C-terminal region is involved in partially inhibiting this interaction, and that in vitroRalA and RalB have an N-terminal calcium-independent and a C-terminal calcium-dependent calmodulin binding domain. Functionally,in vitro Ral-GTP pull-down experiments determined that calmodulin is required for the thrombin-induced activation of Ral in human platelets. We propose that differential binding of calmodulin by RalA and RalB underlies possible functional differences between the two proteins and that calmodulin is involved in the regulation of the activation of Ral-GTPases.

The Ral proteins, RalA and RalB, are small GTPases that are 85% identical and belong to the Ras superfamily of low molecular mass (20 -30 kDa) GTP-binding proteins (1,2). Although biochemically well characterized, the function of Ral proteins in the eukaryotic cell is largely unknown (1,3). They are, however, reported to be involved in signal transduction pathways and have been implicated in the control of cell proliferation and Ras-mediated oncogenic transformation (4,5). Ral guanine nucleotide exchange factors (GEF) 1 RalGDS, Rgl, Rlf, and RGL3 are important downstream effector proteins in Ras signaling pathways (6 -16). Activation of Ral appears to be required for Ras-induced oncogenic growth and morphologic transformation (13,17,18) and induction of DNA synthesis (19). Phospholipase D (PLD) via its multiple lipid second messengers may be one agent that allows Ral to enhance both Rasand Raf-induced cellular transformation (13,20,21). RalA is involved in the tyrosine kinase-and Src-mediated activation of PLD, perhaps through a direct, constitutive association between the N terminus of Ral and PLD1 (13,20,(22)(23)(24). In quiescent rodent fibroblasts, Ral was shown to be sufficient to induce activation of NF-B-dependent gene expression and cyclin D1 transcription (25).
Pathways from Ras to Ral through Ral-GEFs may be selectively regulated by other Ras-like GTPases such as Rap1 (26,27) and TC21 (28). In platelets, Ral and Rap1 were similarly stimulated by platelet agonists ␣-thrombin and platelet-activating factor (29), and both were rapidly activated by elevated levels of calcium (Ca 2ϩ ), which was found to be necessary and sufficient (29,30).
Ral proteins are also intimately involved in endocytosis and exocytosis. RalBP1 regulates endocytosis of epidermal growth factor and insulin receptors (39) by binding to two highly related epsin homology (EH) domain proteins, RalBP1-associated EH domain protein and POB1 (partner of RALBP1) (40,41). It has been suggested (38) that a Ral-RalBP1-POB1 complex transmits signals from receptors to epsin and Eps15, which bind directly to the EH domain of POB1 (39,42) and to the plasma membrane and clathrin adaptor protein complex-2 (43,44), thereby regulating ligand-dependent, receptor-mediated endocytosis (45). Activated Ral may also play a central role in directing sites of exocytosis, because eight specific proteins that comprise the mammalian exocyst complex were found to associate with RalA in a GTP-dependent manner in rat brain (46). The exocyst complex is required for exocytosis and neurite outgrowth, and it localizes to filapodia and neurite growth cones. Therefore, RalA may regulate the integration of receptor and Ca 2ϩ signaling with neurite outgrowth, endocytosis, and directing sites of exocytosis (46). Ral activation is controlled by both Ras-dependent and Rasindependent events (47). Ras-independent activation of Ral occurs in response to elevated levels of Ca 2ϩ (13, 29, 48 -52), and it has been postulated that RalA participates in Ca 2ϩ -dependent intracellular signaling pathways (53). Ral is activated by the Ca 2ϩ ionophore ionomycin, and activation by lysophosphatidic acid or epidermal growth factor can be blocked by a phospholipase C inhibitor (48). The platelet agonists platelet-activating factor, thromboxane A 2 , and ␣-thrombin all activated RalA in platelets (29). The ␣-thrombin-mediated activation of Ral in platelets was inhibited by 1,2-bis(O-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetate (BAPTA-AM)-mediated depletion of intracellular Ca 2ϩ . Ral activation was also stimulated by the thapsigargin-or ionomycin-mediated increase in intracellular Ca 2ϩ , and the increased level of intracellular Ca 2ϩ was sufficient for RalA activation. This suggests that RalA activation was mediated by a common signaling event that involves Ca 2ϩ (29). Calcium stimulated the binding of GTP to RalA/RalB and reduced the binding of GDP to RalA in a dose-dependent manner (50). Calcium may also activate RalA through the binding of calmodulin (CaM) to a region in the C terminus of RalA. Calcium/ calmodulin enhanced by a factor of 3 the GTP binding to RalA isolated from CaM-depleted erythrocyte membranes (51,53). As well, endogenous levels of activated GTP-bound RalA were increased by treatment with ionomycin in rat fibroblasts (48). RalA, but not RalB, contains a putative C-terminal CaM binding domain (BD) that is comprised of basic/hydrophobic residues (amino acids 183-200) and readily forms an amphiphilic helix (53). In vitro experiments with erythrocyte membrane and recombinant RalA demonstrated that RalA bound CaM in a Ca 2ϩdependent manner (53). As well, Ca 2ϩ /CaM dissociated RalA from synaptic vesicle membranes in a Ca 2ϩ -dependent manner (51,54). Because excess EGTA inhibited the dissociation of RalA by CaM, it appears Ca 2ϩ /CaM can dissociate RalA from membranes regardless of whether GTP or GDP is bound to RalA. The GTPase activity of RalA was also reported to be stimulated by Ca 2ϩ /CaM (50). There is also the possibility of cross-talk between signal transduction pathways mediated by Ca 2ϩ /CaM and Ras proteins (53). For example, insulin-like growth factor 1 utilized both the Ras-and Ca 2ϩ -mediated pathways for the coordinate regulation of Ral activity (34).
To determine whether the Ral/CaM interaction occurs in a eukaryotic system, and whether RalB binds CaM (the literature to date only refers to in vitro RalA/CaM interactions), we performed a yeast two-hybrid (Y2H) assay using RalA and RalB as bait to test for interaction with CaM. Results showed that both RalA and RalB interact specifically and directly with CaM in vivo. The speed and degree of positive colony formation suggest that RalA and RalB do not interact identically with CaM. The interaction between the Ral proteins and CaM was confirmed by coimmunoprecipitation experiments in human platelet lysates and shown to be Ca 2ϩ -dependent by in vitro GST and Sepharose fusion protein pull-down experiments with human platelet extracts, and by in vitro binding assays with recombinant radiolabeled proteins. In addition, experiments using C-terminally deleted RalA demonstrated that RalA may have an additional CaM BD to that identified previously (53) in the C terminus of RalA. We also show that both RalA and RalB may contain an N-terminal Ca 2ϩ -independent and a C-terminal Ca 2ϩ -dependent CaM BD and that the C-terminal region of RalB may partially inhibit RalB/CaM interactions. Functionally, we demonstrated that CaM is involved in the thrombininduced activation of RalA and RalB in human platelets. We propose that the differential binding to CaM by RalA and RalB may be the basis for possible functional differences between the two GTPases. We also propose that CaM regulates the activation state of Ral and that Ral may be a downstream effector of many CaM-regulated pathways. The resulting 1661-bp fragment that contained the first 1385 5Ј bp of RIP1 was restricted with AflIII. This generated a 299-bp fragment of RIP1, from bp 1087 (AflIII) to 1385 (BstBI), which was filled in with Klenow fragment and ligated into SmaI-restricted and Klenow-blunted pGEX-4T-2, creating the pGEX-4T-2[RBBD] construct. All cDNA inserts were verified to be in the correct orientation by restriction enzyme analysis and to be in-frame by sequence analysis.
Isolation of GST, GST-RalA, GST-RalB, and GST-RRBD Fusion Proteins-GST and GST fusion proteins were purified as described previously (55). GST-RRBD was purified from AD202 cells after stimulation with isopropyl-1-thio-␤-D-galactopyranoside (55). The purity of the final preparation of the proteins was assessed using SDS-PAGE.
Platelet Washing-Partially purified platelets obtained from the Canadian Blood Services (Winnipeg, Manitoba, Canada) or freshly drawn platelets were gently rocked and treated for 30 min with 0.1 volume of ACD buffer (1.5% citric acid, 2.5% trisodium citrate, 2% glucose) (29). The platelets were centrifuged at 600 ϫ g for 15 min to remove any remaining erythrocytes. The supernatant containing the purified platelets was centrifuged at 1000 ϫ g for 15 min, and the platelet pellets were separated from remaining erythrocytes by washing in HEPES/ Tyrode buffer (HEPES, pH 7.4, 137 mM NaCl, 2.68 mM KCl, 0.42 mM NaH 2 PO 4 , 1.7 mM MgCl 2 , 11.9 mM NaHCO 3 , 5 mM glucose) or buffered saline (10 mM HEPES, pH 7.4, 145 mM NaCl, 5 mM KCl, 10 mM glucose, 1 mM MgSO 4 , 0.5 mM EGTA). The platelets were resuspended in 5-10 ml of HEPES/Tyrode buffer and stored at Ϫ20°C or treated immediately with thrombin after resting at room temperature for 30 min to ensure a quiescent state.
Fractionation of Platelets-Purified platelets were suspended in HEPES/Tyrode buffer or fractionation buffer (20 mM HEPES, pH 7.4, 200 mM KCl, 1 mM MgCl 2 , and 1 mM phenylmethylsulfonyl fluoride) and lysed by sonication. Proteins were centrifuged at 100,000 ϫ g for 2 h at 4°C. The supernatant containing cytosol was used as the source of endogenous CaM. The cytosol was stored at Ϫ80°C until used in the in vitro binding studies. The pellet was resuspended in solubilization buffer (20 mM HEPES, pH 7.4, 200 mM KCl, 1 mM MgCl 2 , 20% glycerol, 0.55% Triton X-100, 1 mM phenylmethylsulfonyl fluoride) and incubated for 1 h at 4°C and centrifuged at 50,000 ϫ g for 30 min. The supernatants containing solubilized particulate proteins were divided into aliquots and stored at Ϫ80°C. To obtain total platelet lysate, purified platelets were suspended and stirred in buffer containing HEPES, pH 7.4, 200 mM KCl, 1 mM MgCl 2 , 0.55% Triton X-100, 20% glycerol, and centrifuged at 100,000 ϫ g for 2 h at 4°C. Total platelet lysate was stored at Ϫ80°C.
GST-Ral Fusion Protein Pull-down Experiments-GST-RalA or GST-RalB (15 g)-agarose beads were incubated with either 4 g/ml purified CaM or 2 mg/ml platelet cytosol proteins for 3 h at 4°C in incubation buffer (20 mM HEPES, pH 7.4, 200 mM KCl, 1 mM MgCl 2 , 0.1% Triton X-100, 20% glycerol). Treatment conditions included either buffer alone, 0.2 mM CaCl 2 , or 5 mM EGTA. In all experiments, GST coupled to GSH-agarose beads was used as control. Following incubation, beads were washed twice using the incubation buffer for each of the treatment conditions and three times with wash buffer (20 mM HEPES, pH 7.4, 200 mM KCl, 1 mM MgCl 2 , 20% glycerol). After addition of 1ϫ Laemmli's sample buffer, the proteins were separated by 13% SDS-PAGE and electrophoretically transferred to PVDF membrane. The blot was probed with anti-CaM antibody (1 g/ml). After incubation with horseradish peroxidase-conjugated secondary antibody (1:5000 dilution), the antigen-antibody complex was visualized using ECL and X-Omat-AR or BIO-MAX-MR film.
In Vitro Transcription/Translation-The pGADT7[CaM] and fulllength and truncated pGBKT7[Ral] constructs, plus empty plasmids as controls (1 g of each), were subjected to in vitro transcription/translation using the TNT Coupled Reticulocyte Lysate System according to manufacturer's protocol. The reaction mix included amino acid mixture minus methionine and [ 35 S]methionine (Ͼ1000 Ci/mmol at 10 mCi/ml). The reactions were incubated for 90 min at 30°C. To test that the correct proteins were translated and in equal amounts, 8 l of reaction mixture was added to 20 l of 1ϫ Laemmli's sample buffer. This was boiled for 2 min, and 8 l of this was subjected to 13 or 15% SDS-PAGE. The gel was subsequently fixed, dried, and exposed on Kodak X-Omat-AR or BIO-MAX-MR film for 8 -16 h at Ϫ70°C, and the radio-activity associated with the appropriate protein was quantitated using a scintillation counter.
In Vitro Binding Assay for CaM and RalA and RalB-[ 35 S]Metlabeled CaM was transcribed and translated in vitro from 1 g of pGADT7[CaM] using the TNT Coupled Reticulate Lysate System. The binding reactions were carried out based on the protocol of Jullien-Flores et al. (33). Briefly, 50 l of Sepharose-CNBr-coupled proteins (CaM BD of RalA (amino acids 183-200) and full-length RalB) and control Sepharose beads as negative control were washed twice in ice-cold in vitro binding buffer (20 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 200 mM NaCl, 0.5% Nonidet P-40, and 1 mM 4-[(2-aminoethyl)]-benzenesulfonyl fluoride (AEBSF)) and incubated overnight at 4°C with equal amounts (ϳ10 l) of in vitro transcribed/translated 35 S-CaM in 100 l of binding buffer. After sedimentation of the beads, the supernatant was removed, and the beads were washed three times with ice-cold binding buffer containing 1 mM dithiothreitol. The bound proteins were recovered by boiling the beads in 1ϫ Laemmli's sample buffer and separated by 13% SDS-PAGE. The gel was fixed for 1 h in 50% methanol, 10% acetic acid, and 40% water, washed twice in doubledistilled water for 15 min each, treated for 1 h with 1 M sodium salicylate, pH 7.0, and soaked in pre-drying buffer (7% methanol, 7% acetic acid, 1% glycerol) for 10 min. The gel was dried, and the presence of 35 S-CaM was detected by autoradiography at Ϫ70°C for 1-7 days. In the reverse reaction, the in vitro binding reaction was repeated with in vitro transcribed/translated [ 35 S]methionine-labeled full-length and truncated Ral constructs. The labeled proteins were incubated with 50 l of CaM-Sepharose or blank Sepharose 4B beads as control. In all assays, the binding buffer was used with no additions or with 0.5 mM CaCl 2 or 5 mM EGTA/EDTA added.
Coimmunoprecipitation-Freshly drawn and outdated human platelets were prepared as above and divided into 1-ml aliquots. Fresh platelets were treated with 0.2 units/ml thrombin for 60 s at 37°C without stirring. The outdated platelets were lysed 1:1 v/v with ice-cold 2ϫ platelet RIPA buffer (2% Triton X-100, 2% sodium deoxycholate, 0.2% SDS, 316 mM NaCl, 2 mM EGTA, 20 mM Tris, pH 7.6) containing protease inhibitors at a final concentration of 1 mM AEBSF, 10 g/ml aprotinin, 10 g/ml leupeptin, and 50 mM benzamidine, plus 1 mM sodium orthovanadate. Fresh platelets were lysed by adding 1:1 v/v of 2ϫ platelet RIPA buffer minus Triton X-100, deoxycholate, and SDS but with protease inhibitors added. The mixtures were put on ice for 30 min (RIPA-lysed) or sonicated (fresh platelets) at 4°C. The lysates were cleared by centrifugation (17,000 ϫ g for 60 min), and the supernatants were precleared with 20 l of protein A/G Plus-agarose and 1 g of goat anti-rabbit IgG (RIPA-lysed) or 0.2 g of rabbit anti-mouse IgG1 (sonication-lysed) with gentle rocking at 4°C for 30 -60 min. The supernatants were cleared by centrifugation (16,000 ϫ g for 1 min), incubated with 2 g/ml anti-RalB polyclonal (RIPA-lysed) or 5 g/ml anti-CaM monoclonal (sonicated) antibody, and rocked at 4°C for 1 h. Control platelet lysates were incubated with 2 g of rabbit IgG (RIPA-lysed) or 1 g of IgG1 (sonicated). Protein A/G Plus-agarose (30 l) was then added, and the mixtures were rocked at 4°C for 30 -60 min. The beads were collected by centrifugation (16,000 ϫ g for 1 min) and washed four times in 1ϫ platelet RIPA buffer (RalB antibody) or TN buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl) (CaM antibody). The beads were resuspended in 25 l of Laemmli's buffer, boiled for 3 min, and centrifuged at 17,000 ϫ g for 4 min, and the proteins were separated by 13 (RalB immunoprecipitate) or 15% (CaM immunoprecipitate) SDS-PAGE. Proteins were transferred to PVDF membranes and probed with anti-CaM (1 g/ml) monoclonal (RalB immunoprecipitate) or anti-RalB (1:125) polyclonal (CaM immunoprecipitate) antibody. The membrane containing CaM-precipitated proteins (sonicated) was stripped and reprobed with anti-CaM monoclonal antibody to verify the presence of CaM.
Detection of GTP-bound Ral Using GST-RIP1 Ral BD-Experimental procedures were based on the protocol of Wolthuis et al. (29). Freshly drawn platelets were divided into 500-l aliquots, and stimulation with 0.2 units/ml thrombin was performed at 37°C without stirring for 10 and 90 s. W7 and W5 (50 M) were added 10 min before agonist treatment. Platelets were lysed (2:1 v/v) in ice-cold 3ϫ Ral buffer (final concentration: 10% glycerol, 1% Nonidet P-40, 50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 2.5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 1 mM AEBSF, 1 M leupeptin, 0.1 M aprotinin, 25 M pepstatin A, 1 mM benzamidine) for 30 min at 4°C with gentle rocking. Lysates were clarified by centrifugation at 16,000 ϫ g at 4°C for 10 min, and 0.75 ml of the supernatant of each sample was incubated with 50 l of GST-RRBD precoupled to glutathione beads. Samples were incubated for 1 h on a rocker at 4°C. Beads were collected by centrifugation (9,000 rpm for 1 min) and washed 4 times in Ral buffer. Beads were resuspended in 50 l of 1ϫ Laemmli buffer and boiled, and 25 l of this was subjected to SDS-PAGE in duplicate, and Western blotting was performed with anti-RalA monoclonal (1:5000) and anti-RalB polyclonal (1:125) antibodies.
Yeast Two-hybrid Assay-CLONTECH's Matchmaker Two-hybrid System 3 protocol was followed. AH109 yeast cells were transfected with RalA or RalB cDNAs or their deletion constructs subcloned into pGBKT7, plus CaM cDNA subcloned into pGADT7 expression plasmid. Following small and large scale yeast transformation protocols, using simultaneous and sequential transformation procedures, the transfected AH109 cells were spread onto minimal SD agar plates lacking histidine, adenine, leucine, and tryptophan (SD/ϪHALT), and with 5-bromo-4-chloro-3-indolyl ␣-D-galactopyranoside (X-␣-Gal) spread onto the surface of each plate. The plates were incubated at 30°C for 2-10 days. Any positive blue colonies were restreaked at least three times onto fresh SD/ϪHALT/ϩX-␣-Gal plates to ensure propagation of correct phenotype. As positive controls, AH109 was transfected with (i) pCL1, which contains the complete GAL4 transcription factor, and spread onto SD/ϪLeu/ϩX-␣-Gal plates; and (ii) pGBKT7[murine p53] and pGADT7[SV40 large T-antigen], which were spread onto SD/ϪTrp/ ϪLeu/ϩX-␣-Gal plates. As negative controls, AH109 was transformed with (i) pGBKT7 Yeast Mating-Small and large scale yeast matings were performed as per CLONTECH's protocols. AH109 cells transfected with full-length and truncated RalA and RalB cDNAs subcloned into pGBKT7 were mated with Y187 yeast cells transfected with pGADT7[CaM]. Mated yeast cells were spread onto SD/ϪHALT/ϩX-␣-Gal plates. Any positive blue colonies that grew were restreaked at least 3 times onto fresh SD/ϪHALT/ϩX-␣-Gal plates to ensure correct phenotype.
All transfected yeast were tested for expression of appropriate fusion proteins by Western analysis. Yeast proteins were extracted by CLON-TECH's urea/SDS method, separated by 13% SDS-PAGE, and probed with appropriate antibodies (anti-c-Myc for RalA/RalB and anti-hemagglutinin for CaM). Results were also used to determine qualitatively that similar amounts of protein were expressed.
Molecular Biology Methods-Restriction enzyme techniques, DNA extraction from agarose gels, dephosphorylation reactions, DNA polymerase 1 (Klenow) reactions, DNA ligation, plasmid transformation of bacteria, and plasmid preparations were all performed as per standard molecular biological procedures, following the manufacturers' protocols.

Ca 2ϩ -dependent in Vitro Binding of Endogenous CaM by
GST-RalA and GST-RalB-It has been shown previously (53) that RalA has a C-terminal CaM BD. To determine, for the first time, whether both RalB and RalA interact with CaM, an in vitro GST fusion protein pull-down experiment and Western blotting with anti-CaM antibody were performed. GST-RalA (Fig. 1, lane 1) and GST-RalB (Fig. 1, lane 4) fusion proteins were found to be associated with endogenous CaM from human platelet cytosol fractions (Fig. 1, lane 10). This interaction was Ca 2ϩ -dependent, because the addition of 0.2 mM CaCl 2 (Fig. 1,  lanes 3 and 6) enhanced the interaction, and 5 mM EGTA (Fig.  1, lanes 2 and 5) almost eliminated the binding of Ral to CaM. There was no binding of CaM to GST under any of these conditions (Fig. 1, lanes 7-9). Therefore, both RalA and RalB interacted with endogenous CaM in this in vitro system in a Ca 2ϩ -dependent manner. That identical results were obtained using purified bovine CaM (Fig. 2) suggests the binding of Ral to CaM is direct. This appears to be the first report of RalB interacting with CaM.
Ca 2ϩ -dependent in Vitro Binding of Endogenous RalA and RalB by CaM-Sepharose-CaM-Sepharose was associated with RalA and RalB extracted from human platelet-solubilized particulate fractions, as detected by an in vitro CaM affinity binding assay and Western blotting with anti-RalA and anti-RalB (Fig. 3, A and B, lane 2, respectively) antibodies. These interactions were Ca 2ϩ -dependent because 5 mM EGTA almost eliminated the binding between CaM and RalA and RalB (Fig. 3, A and B, lane 3, respectively) that was detected in the particulate fractions, as well as in those fractions supplemented with 0.2 mM CaCl 2 (Fig. 3, A and B, lane 4, respectively). As negative controls, Sepharose beads did not pull down RalA or RalB (Fig.  3, A and B, lane 1), and Western blot analysis did not detect Rab5A or Rab6, despite being present in high amounts in the platelet particulate fraction (data not shown). These results suggest that CaM interacts specifically with endogenous platelet RalA and RalB in a Ca 2ϩ -dependent manner.
Recombinant RalA/B and CaM Interact Specifically in Vitro in a Ca 2ϩ -dependent Manner-The interactions between CaM and the Ral proteins were also tested by an in vitro binding reaction (33). Autoradiography showed that [ 35 S]Met-labeled RalA interacted with CaM-Sepharose in the presence of 0.5 mM CaCl 2 (Fig. 4A, lane 1) at a significantly greater degree than in the presence of 5 mM EDTA (Fig. 4A, lane 2). In the reverse reaction, Sepharose-CNBr-coupled CaM BD of RalA (Fig. 4B, lane 2) and recombinant RalB (Fig. 4B, lane 3) formed complexes with 35 S-CaM. This Ral/CaM interaction was Ca 2ϩ -dependent. CaM bound to Sepharose-CNBr-coupled CaM BD of RalA (Fig. 4C, lane 2) and recombinant RalB (Fig. 4C, lane 3) in the presence of 0.5 mM CaCl 2 but not in the presence of 5 mM EDTA (Fig. 4C, lanes 5 and 6, respectively). The interactions were specific because Sepharose-CNBr control beads did not bind CaM (Fig. 4, A-C, lane 1). These results indicate that recombinant Ral and CaM proteins interact in a Ca 2ϩ -dependent manner.
Endogenous CaM and Ral Coimmunoprecipitate in Human Platelets-Coimmunoprecipitation experiments were performed to establish whether endogenous Ral and CaM form complexes in platelets. CaM coprecipitated from total platelet extract derived from outdated platelets with anti-RalB but not preimmune rabbit IgG (Fig. 5A) antibody. In the reverse reaction to show that Ral-CaM exists as a complex, initial experiments demonstrated that CaM could not be immunoprecipitated using anti-CaM antibody from detergent-solubilized platelets. We have shown previously (55) that in platelets, RalB but not RalA is present in both the particulate and cytosol fraction. Thus, cytosol was used to immunoprecipitate RalB-CaM complex using anti-CaM antibody. RalB coprecipitated from platelet cytosol fraction derived from thrombin-stimu-  3, 6, and 9, respectively) or 5 mM EGTA (lanes 2, 5, and 8, respectively). Platelet cytosol was also probed with anti-CaM antibody (lane 10). These experiments were repeated at least five times and gave identical results. lated (Fig. 5B) or -unstimulated (data not shown) fresh platelets to a much greater degree with anti-CaM compared with preimmune IgG antibody. Reprobing of the membrane with anti-CaM antibody showed that CaM was present in the immunoprecipitate (data not shown). The coimmunoprecipitation results provide further evidence that Ral-CaM interact and that active and inactive Ral form complexes with CaM in platelets.
Calmodulin Interacts Specifically with RalA and RalB in Vivo-To relate these findings to mammalian cells, it was also important to demonstrate this Ral/CaM interaction in vivo in a eukaryotic system. Therefore, a Y2H assay was employed, which demonstrated, for the first time, specific interaction of the Ral proteins with CaM in vivo. Positive blue colonies, caused by the protein-protein-induced activation of the yeast MEL1 gene and subsequent synthesis of ␣-galactosidase and digestion of X-␣-Gal, resulted when AH109 cells were transfected with pGADT7[CaM] and pGBKT7[RalA] or pGBKT7[RalB] (Fig. 6B, RalA ϩ CaM and RalB ϩ CaM). Positive controls (i) pCL1 (Fig. 6B, pCL1) and (ii) pGBKT7[murine p53] plus pGADT7[SV40 large T-antigen] (Fig. 6B, T ϩ 53) also produced blue colonies, whereas negative controls (i) pGBKT7[lamin C] plus pGADT7[T-antigen] (Fig.  6B, T ϩ Lam) and (ii) pGBKT7[lamin C] plus pGADT7[CaM] (data not shown) and the Ral (Fig. 6B, RalA and RalB) and CaM (Fig. 6B, CaM)  Both RalA and RalB interacted with CaM to induce synthesis of ␣-galactosidase (data not shown). The blue phenotype in both the yeast transfection and mating experiments persisted upon restreaking 3-4 times on high stringency drop-out media. In all cases, the positive RalA/CaM colonies grew more quickly and developed a more intense blue coloration than the RalB/ CaM colonies, even though Western blotting showed, qualitatively, similar levels of protein were expressed by the yeast (data not shown). These results demonstrate that both RalA and RalB directly bind CaM in vivo in a eukaryotic system and that RalA may bind CaM more readily than RalB.
CaM Interacts with C-terminally Truncated Ral Proteins in a Ca 2ϩ -dependent Manner-A CaM target data base (calcium.

FIG. 2. GST-RalA and GST-RalB bind to purified bovine brain CaM in a Ca 2؉ -dependent manner.
In a GST fusion protein pull-down experiment, 15 g of agarose-GST-RalA (lanes 1-3) and -RalB beads (lanes 4 -6) were incubated with purified CaM under various conditions as described under "Experimental Procedures." Bound proteins were separated by 13% SDS-PAGE, transferred to PVDF membrane, and probed with anti-CaM monoclonal antibody to monitor the interaction between CaM and GST-RalA (lane 1) and GST-RalB (lane 4) in incubation buffer alone and in the presence of 0.2 mM CaCl 2 (lanes 3 and 6 respectively) or 5 mM EGTA (lanes 2 and 5 respectively). Bovine brain CaM (0.5 g) was also probed with the anti -CaM antibody (lane 7). These experiments were repeated at least five times and gave identical results. oci.utoronto.ca/) suggests that the N-terminal regions of both RalA and RalB have a propensity to form ␣-helical wheels and thus may be CaM binding domains. We therefore tested whether removal of the C-terminal region equivalent to that proposed to contain the RalA CaM BD (53) would eliminate CaM binding to RalA and RalB. Fig. 6A shows a schematic of full-length and truncated Ral constructs used in these experiments. In the yeast systems, RalA-(1-549) (Fig. 6B, RalA ⌬C-T ), which has the C-terminal CaM BD deleted, RalB-(1-482) (data not shown), which has the region equivalent to the C-terminal CaM BD of RalA deleted, and RalB-(100 -482) (Fig. 6B, RalB ⌬N-T/⌬C-T ), which has both the putative N-terminal and C-terminal CaM BDs deleted, appeared to bind CaM but more weakly than full-length RalA and RalB. This conclusion was based solely on the subjective determination of speed of onset and degree of blue coloration of the various colonies. In support of these in vivo results, in vitro binding assays showed that Sepharose-CaM bound 35 S-RalA-(1-549) (Fig. 7A, lane 2) and 35 S-RalB-(1-482) (Fig. 7B, lane 2), as well as full-length 35 S-RalA and 35 S-RalB (Fig. 7, A and B, lane 1, respectively) in the presence of 0.5 mM CaCl 2 , whereas the control Sepharose beads did not bind these proteins (Fig. 7, A, lanes 3 and 4, and B,  lanes 4 and 5). In fact, the C-terminally truncated 35 S-RalB product appeared to bind Sepharose-CaM much more readily than full-length 35 S-RalB (Fig. 7B, compare lanes 1 and 2). However, Sepharose-CaM did not significantly bind more of the doubly truncated RalB product ( 35 S-RalB-(100 -482)) (Fig. 7B,  lane 3) over that of control Sepharose beads (Fig. 7B, lane 6). In all cases, the CaM/Ral interaction was also Ca 2ϩ -dependent because 5 mM EGTA eliminated the binding detected in the presence of 0.5 mM CaCl 2 (data not shown). These results suggest that there is more than one CaM BD in RalA and probably RalB. Because the double N-and C-terminally truncated RalB construct failed to bind CaM in vitro and very weakly in vivo, it appears that RalB may indeed have an N-terminal and a C-terminal CaM BD or, alternatively, that both regions are necessary for binding.
CaM Is Required for the Thrombin-induced Activation of RalA and RalB-We used the RRBD to determine whether CaM plays a role in the activation of Rals. Initially, to check specificity of our GST-RRBD, 1 ml of thrombin-stimulated (75 s) and -unstimulated freshly drawn human platelets were subject to GST-RRBD and control GST pull-down experiments (as described under "Experimental Procedures"). The Ral BD motif of RIP1 selectively binds GTP-bound but not GDP-bound Ral (29). Only thrombin-stimulated platelets showed significant amounts of Ral-GTP, with unstimulated platelets showing negligible amounts. The reaction was specific because GST control beads did not bind activated Ral in either unstimulated or thrombin-stimulated platelets (data not shown). The next step was to determine whether CaM was involved in the activation of Ral. Thrombin led to full activation of RalA (Fig. 9, upper panel) and RalB (Fig. 9, lower panel) within 10 s. This activation was eliminated in the presence of the CaM inhibitor W7⅐HCl but not W5⅐HCl. W5 was used at a low non-CaMinhibiting concentration to show that CaM inhibitors do not autonomously prevent Ral activation. These results suggest that CaM functions in the activation and regulation of RalA and RalB. DISCUSSION It has been reported previously (7,13,29,48,49,51,52) that elevated levels of Ca 2ϩ activate RalA and that CaM binds RalA in vitro (51,53), leading to the proposal that RalA is associated with Ca 2ϩ /CaM-dependent intracellular signaling pathways (53). As well, platelet agonists (e.g. platelet-activating factor, thromboxane A 2 , and ␣-thrombin) activate RalA in platelets in a Ca 2ϩ -dependent manner, suggesting that RalA activation is mediated by a common signaling event that may involve Ca 2ϩ (29). We propose that CaM is an essential component in this process. We have shown, for what appears to be the first time, by coimmunoprecipitation in human platelets and yeast twohybrid experiments, that RalB binds CaM and that both RalA and RalB interact specifically and directly with CaM in vivo in a eukaryotic system. These results were confirmed and shown to be Ca 2ϩ -dependent by in vitro binding assays, as well as GST and Sepharose fusion protein pull-down assays in human platelet lysates. Results from the Y2H assays appear to show that RalA binds CaM more readily than RalB. This was based solely on the qualitative finding that positive blue RalA/CaM colonies formed more quickly and had a more intense blue coloration than RalB/CaM colonies, even though Western analysis showed similar protein expression levels. We speculate that this differential binding of CaM by the Ral proteins may determine functional differences between RalA and RalB, as related to Ca 2ϩ /CaM signaling pathways. Further study is needed to examine these preliminary findings of differential binding and to determine the functional significance.
We have shown that in human platelets, thrombin induces the activation of both RalA and RalB in a CaM-dependent manner. This would suggest that the Ca 2ϩ -dependent platelet agonist-induced (29) and Ca 2ϩ -induced (29,30) activation of RalA reported previously may in fact have been both CaM-and Ca 2ϩ -dependent. It is likely that Ral is activated by many of the CaM-regulated intracellular pathways by forming a complex with CaM and its associated proteins. Several studies propose Ca 2ϩ is involved in Ral activation by various mechanisms (4,29,50). We speculate that CaM may be required for some, if not all, of these Ca 2ϩ -dependent regulatory pathways. Ca 2ϩ /CaM induces GTP binding to RalA in human erythrocyte plasma membrane (48,51,53), and the cycling of RalA in synaptic vesicle membranes by CaM is Ca 2ϩ -dependent (50,54). Further study is needed to determine exactly which of these CaMregulated pathways are involved with Ral, the mechanism by which CaM induces Ral activation, and the results of such CaM-induced Ral activation on downstream effectors and their pathways. Our results demonstrating the thrombin-induced, CaM-dependent activation of RalA and RalB in platelets, plus results cited from the literature, suggest that Ca 2ϩ /CaM activates both RalA and RalB and that the signaling pathways of Ca 2ϩ /CaM and RalA (54) and RalB are directly linked.
The regulation of Ral function by CaM appears to be complex. Initially, the CaM BD in RalA has been shown to be at the C terminus (53). Because the two Ral proteins differ mainly at the C terminus, we wished to determine whether there was a CaM BD in another region of RalB and perhaps RalA. The sequences of RalA and RalB were analyzed using the CaM target data base (calcium.oci.utoronto.ca/). This revealed the presence of a potential N-terminal CaM BD in RalB (average propensity for ␣-helix formation, 1.052). Examination of RalA showed a second high scoring putative N-terminal CaM BD (average propensity 0.957) when the C-terminal 30 residues containing the predicted CaM BD (53) were removed to allow for a pattern search to identify any additional potential CaM BDs. RalA and RalB have almost identical values for various parameters obtained from the analysis of the predicted Nterminal CaM-binding sites. The mean hydrophobicity (Ϫ0.125, RalA, and 0.135, RalB) and hydrophobic moments (0.302, RalA, and 0.628, RalB) for the N-terminal regions of both proteins are within the range of values of most CaM BDs (53). The major difference between the putative N-terminal CaM BDs and the predicted C-terminal CaM BD of RalA is that the latter forms a hydrophilic ␣-helical wheel (53), and the N-terminal regions form hydrophobic ␣-helical wheels. The interaction of CaM with its target proteins is predominantly hydrophobic (53). Both N-and C-terminal regions carry a net positive charge (ϩ3 for N-terminal and ϩ9 for C-terminal) because binding of CaM with target proteins also involves strong electrostatic interactions (53). Therefore, the possibility of more than one CaM BD being present in the Ral proteins was FIG. 6. Calmodulin interacts specifically with RalA and RalB in vivo. A, full-length and truncated RalA and RalB constructs. Arrowheads denote cleavage sites, and numbers represent nucleotide base pairs. B, calmodulin interacts specifically with various RalA and RalB constructs in vivo. RalA and RalB cDNAs and their deletion constructs were subcloned into GAL4 BD expression plasmid pG-BKT7, and CaM cDNA was subcloned into GAL4 AD expression plasmid pGADT7. To identify specific protein-protein interactions, yeast two-hybrid and yeast mating assays were performed as described under "Experimental Procedures." Colonies were monitored for blue coloration caused by the direct interaction of the GAL4 BD and AD fusion proteins. examined. To test what effect on CaM binding removal of the C-terminal regions of the Ral proteins would have, the appropriate regions were deleted from RalA and RalB, and the subsequent truncated proteins were tested for interaction with CaM. Results demonstrated that RalA-(1-549), which lacks the predicted CaM BD (53), and RalB-(1-482), which lacks the region equivalent to that of the CaM BD of RalA, interacted specifically and directly with CaM in the Y2H and in vitro binding assays. The in vitro results showed the interaction to be Ca 2ϩ -dependent. It is realized that deleting large portions of a protein may alter its three-dimensional structure and, hence, its ability to interact normally with other proteins. However, in vitro binding assays using N-and C-terminal halves of each Ral protein suggest that their C-terminal halves contains a Ca 2ϩdependent CaM BD motif, whereas their N-terminal halves contain a Ca 2ϩ -independent CaM BD. Further experiments (e.g. site-directed mutagenesis) are underway to define the CaM binding sequence in the N-and C-terminal regions of Rals.
Results therefore indicate that there is an additional CaM BD in RalA from that described in the literature (53) and at least two CaM BDs in RalB. This is supported by the in vitro binding assay results with the N-and C-terminally truncated RalB construct which failed to bind CaM. The in vitro results appear to show C-terminally truncated RalB binds CaM much more readily than full-length RalB and that the deleted portion contains the Ca 2ϩ -dependent CaM BD. These C-terminal 40 or so amino acids of RalB may normally be involved in inhibiting CaM/RalB interactions. This may partly explain the differential binding of CaM by RalA and RalB. We speculate that in the cell, the putative RalB CaM-binding inhibitory region is masked or inhibited by another molecule or by a change in conformation. Because this inhibitory region contains an apparent Ca 2ϩ -dependent CaM BD, the suggested molecular or conformational regulation may be controlled by Ca 2ϩ . We propose that the inhibitory area is masked at high intracellular Ca 2ϩ concentrations and is open and fully active at low Ca 2ϩ levels. In addition, optimal binding of CaM to RalA and RalB may require both Ca 2ϩ -dependent and Ca 2ϩ -independent CaM BDs. For instance, once CaM binds to the C-terminal CaM BD, the N-terminal domain may bind CaM with greater affinity.
It will be necessary to demonstrate Ral/CaM interactions in mammalian cells and to determine how Ca 2ϩ , CaM, and other agents (e.g. agonists of platelets and cell transformation) affect the activation of Ral in cells. Identification of additional Ralinteracting proteins is required to discover more CaM-regulated pathways that activate Ral. Examination of suspected  4 -6, respectively). In both A and B, bound proteins were recovered from beads by boiling in 1ϫ Laemmli's buffer and subjected to 13% SDS-PAGE. The gel was vacuum-dried, and autoradiography was used to visualize radioactivity associated with the proteins. These experiments were repeated at least three times and gave identical results.  , lanes 1, 3, 5, and 7), and RalB-(317-621) (lower panel, lanes 2, 4, 6, and 8) were incubated with 50 l of CaM-Sepharose 4B beads in the presence of 0.5 mM CaCl 2 or 5 mM EDTA as indicated or incubated with 50 l of control Sepharose 4B beads plus 0.5 mM CaCl 2 as a negative control (blank beads). In vitro transcribed/translated [ 35 S]Met-labeled Ral constructs (TNT reaction) were run as positive controls. Bound proteins were recovered from beads by boiling in 1ϫ Laemmli's buffer and separated by 15% SDS-PAGE. The gel was vacuum-dried, and autoradiography was used to visualize radioactivity associated with the proteins. These experiments were repeated at least three times and gave identical results.
FIG. 9. Calmodulin is required for the thrombin-induced activation of RalA and RalB. Freshly drawn human platelets were prepared as described under "Experimental Procedures," divided into 500-l aliquots, and incubated for 30 min at 37°C prior to thrombin treatment (0.2 units/ml). Ten minutes before thrombin treatment, the indicated aliquots were treated with either 50 M W7 or W5. After thrombin treatments for 10 or 90 s, the platelets were lysed in 3ϫ Ral buffer and incubated with 60 l of GST-RRBD precoupled to glutathione beads to recover GTP-bound Ral. Beads were washed 4 times in Ral buffer; proteins were boiled off in Laemmli buffer, and the collected Ral-GTP was identified by Western analysis with anti-RalA (1:5000, upper panel) or anti-RalB (1:125, lower panel) antibodies. These experiments were repeated at least three times and gave identical results. differences in CaM binding between the Ral proteins warrants further attention and also whether the guanine nucleotide status of Ral affects its CaM binding affinity. How Ral-GTPases regulate, or are regulated themselves, by Ca 2ϩ /CaM-dependent pathways will shed light on important cellular processes in health and disease.