Activation of R-Ras by Ras-Guanine Nucleotide-releasing Factor*

Ras-GRF/CDC25Mm, mSos, and C3G have been identified as guanine nucleotide-releasing factors for Ras family proteins. We investigated in this study the guanine nucleotide-releasing activities of Ras-GRF, mSos, and C3G toward R-Ras, which shows high sequence similarity to Ras. Ras-GRF markedly stimulated the dissociation of GDP from R-Ras, and C3G also promoted the release of R-Ras-bound GDP. Under the same conditions, mSos little affected the reaction. When Ras-GRF and R-Ras were coexpressed in COS7 cells, the remarkable accumulation of the active GTP-bound form of R-Ras was observed. C3G also increased active R-Ras in COS7 cells, while mSos did not give any effect. These results indicated that Ras-GRF and C3G could activate R-Ras. Furthermore, the activation of R-Ras by Ras-GRF was enhanced when cells were treated with ionomycin, which is known to increase the intracellular calcium concentration. The examination of tissue distribution of R-Ras, Ras-GRF, and mSos by the reverse transcription-polymerase chain reaction revealed that Ras-GRF was expressed only in brain and testis, whereas R-Ras, C3G, and mSos were expressed rather ubiquitously. These findings raise the possibility that R-Ras is activated by Ras-GRF in brain and testis, and by C3G in other tissues, respectively.

In contrast to activated Ras, the activated form of R-Ras does not transform Rat1 fibroblastic cells (1). Rey et al. (7) stated that R-Ras did not induce DNA synthesis in Swiss3T3 cells or the differentiation of PC12 cells. However, two further reports demonstrated that activated R-Ras transforms NIH3T3 cells and the transformant forms tumors in athymic nude mice (8,9). Since the putative effector domain of R-Ras is highly homologous to that of Ras, R-Ras binds to c-Raf-1 kinase and activates the mitogen-activated protein kinase pathway leading to the induction of c-fos (7,8,10). Besides the roles in the stimulation of the proliferation of cells, R-Ras may play roles in some cell biological processes other than those of Ras. It was described that the carboxyl-terminal domain of R-Ras interacts with Bcl-2, which is a key molecule in the control of apoptosis, in a GTP-independent manner (11). The activated form of R-Ras promotes apoptotic cell death induced by growth factor deprivation through a Bcl-2 suppressible mechanism (12). Zhang et al. (13) stated that the active form of R-Ras stimulates cell adhesion through the activation of integrins.
Following the identification of Ras-GRF and mSos as mammalian GRFs for Ras, we isolated C3G as a Crk SH3-binding GRF (14). C3G has a sequence similarity to the catalytic domains of Ras-GRF and mSos. The fact that C3G complemented a temperature-sensitive mutation of Saccharomyces cerevisiae CDC25 gene indicates that C3G can activate RAS in budding yeast cells (14). However, additional biochemical and cell biological analyses revealed that C3G activates Rap1 much more efficiently than does Ras (15). A GRF for Ral was also described (16); however, a GRF specific to R-Ras has not yet been identified.
Considering the high sequence similarity between Ras and R-Ras, we examined the effects of the known GRFs for the Ras family proteins, Ras-GRF, mSos, and C3G on the guanine nucleotide release from R-Ras. The results described in this study indicate that Ras-GRF and C3G can stimulate the nucleotide exchange of R-Ras and raise the possibility that these factors may be activators for R-Ras.

EXPERIMENTAL PROCEDURES
Low Molecular Weight GTP-binding Proteins-Ha-Ras and R-Ras expression plasmids (17), generous gifts of Dr. K. Kaibuchi, directed the expression of Ha-Ras and R-Ras as fusion proteins with glutathione S-transferase (GST). GST fusion proteins were purified by using glutathione-Sepharose 4B (Pharmacia Biotech Inc.) and digested with thrombin (Sigma). Rap1B protein produced in Sf9 cells using a baculovirus vector (18) and Escherichia coli producing the full-length N-Ras protein were kind gifts from Drs. Y. Takai and A. Wittinghofer, respectively.
pEBG-R-Ras, a mammalian GST fusion R-Ras expression vector derived from pEBG (19), was constructed as follows. An entire coding region of R-Ras was amplified by polymerase chain reaction (PCR), and the product was inserted into the BamHI site of pEBG. The Ha-Ras expression plasmid SR␣Ras was described previously (20).
GRFs-The full-length cDNA of Ras-GRF was kindly provided by Dr. L. Feig. The SalI/ClaI fragment of Ras-GRF covering the entire coding region was inserted into XhoI site of an expression vector, pCAGGS (21), which yielded pCAGGS-Ras-GRF. The membrane targeting farnesylation signal, CAAX box of the c-Ki-ras2 gene, was amplified by PCR with primers Ki-Ras CAAX-5Ј and Ki-Ras CAAX-3Ј (15). The cDNA of the carboxyl-terminal region of Ras-GRF was amplified by PCR with primers Ras-GRF-5Ј (5Ј-GGCAAGCTTGTGTCATCAGAT-3Ј) and Ras-GRF-3Ј (5Ј-GGCTCTAGATGTGGGGAGTTTTGG-3Ј). In the product, Ras-GRF-C term, the authentic termination codon of Ras-GRF was replaced with an XbaI site, to which the CAAX box sequence amplified as described above was fused. A HindIII-EcoRI fragment of the fused sequence covering the carboxyl-terminal region of Ras-GRF with the CAAX box was used to replace the corresponding part of pCAGGS-Ras-GRF to generate pCAGGS-Ras-GRF-F (F stands for farnesylation signal). pCAGGS-mSos, pCAGGS-mSos-F, pCAGGS-C3G, and pCAGGS-C3G-F, which express mSos and C3G with or without the farnesylation signal, were described previously (15).
Ras-GRF and C3G proteins for in vitro experiments were described previously (15). The catalytic domain of mSos (22) was amplified by PCR using a primer mSos-N (5Ј-GGCATCGATCAGCTGAAGAGAA-AAAC-3Ј), corresponding to amino acid residues 559 -565 with an artificial ClaI site, and a primer mSos-C (5Ј-TCCTCGGTCTTGGATTT-GATT-3Ј), corresponding to amino acids 1070 -1077 with an artificial NotI site. An amplified mSos fragment was ligated to ClaI/NotIdigested pEBG, to yield pEBG-mSos. To obtain a mSos protein, 1 g of pEBG-mSos was introduced into COS7 cells (1.0 ϫ 10 6 cells per 100-mm diameter dish) that had been cultured in Dulbecco's modified Eagle's medium (Nissui, Tokyo) supplemented with 10% fetal calf serum. After 48 h, the cells from six dishes were harvested and suspended in 500 l of a solution containing 20 mM Tris-HCl (pH 7.6), 150 mM NaCl, 10% glycerol, 1 g/ml leupeptin, and 1 g/ml antipain and disrupted by freezing and thawing. The lysate was centrifuged at 10,000 ϫ g for 5 min, and the supernatant was saved and used as the mSos protein. The pEBG vector was similarly transfected into COS7 cells, and the lysate was used as a control.
The sequences of all the amplified fragments described above were confirmed after subcloning the fragments into pUC18.
Measurement of Guanine Nucleotide Exchange Reaction-The nucleotide exchange reaction was carried out as described previously (15). Briefly, 20 l of a reaction buffer containing 2.0 pmol of each low molecular weight GTP-binding protein-[ 3 H]GDP complex, an appropriate amount of GRF, and an excess amount of nonradioactive GTP were incubated at 30°C for 20 min. The activities of exchange factors were measured as a decrease in the amount of GTP-binding protein-[ 3 H]GDP complex, which was quantitated by the nitrocellulose filter method.

Analysis of Guanine Nucleotides Bound to R-Ras and H-Ras in Intact
Cells-COS7 cells (1.5 ϫ 10 5 per 35-mm diameter dish) were transfected with the various combinations of expression vectors for Ha-Ras, R-Ras, and GRFs (0.3 g) (15). After 48 h, guanine nucleotides bound to Ha-Ras or R-Ras were analyzed as described previously (15,23). Briefly, cells were labeled with 0.05 mCi/ml [ 32 P]orthophosphate (Amersham Corp.) in a serum-free medium for 4 h, then the cells were lysed in a lysis buffer. Ras was immunoprecipitated by Y13-259 anti-Ras monoclonal antibody, and GST-R-Ras was recovered by using glutathione-Sepharose. Guanine nucleotides bound to Ras and R-Ras were separated by polyethyleneimine thin layer chromatography and quantitated by the BAS 2000 system (FUJI Film, Tokyo). The treatment of cells with ionomycin was carried out as described by Farnsworth et al. (24). Thirty-two hours after transfection of the plasmids, cells were incubated in a serum-free medium for 12 h and then labeled with [ 32 P]orthophosphate for 4 h. Cells were treated with 10 M ionomycin for 5 min, then R-Ras-bound nucleotides were analyzed.

Ras-GRF Stimulates GDP Release from R-Ras-
We measured the substrate specificity of Ras-GRF by using various Ras family proteins. As expected Ras-GRF stimulated the dissociation of GDP from Ha-Ras and N-Ras in a dose-dependent manner (Fig. 1, left). Ras-GRF also stimulated the dissociation of GDP from R-Ras. We previously reported that Ras-GRF did not stimulate the release of GDP from RalA, Rab3A, or RhoA (15).
We next examined the substrate specificity of mSos, another GRF for Ras. Under the same conditions, mSos stimulated the dissociation of GDP from Ha-Ras and N-Ras, but not from R-Ras in vitro (Fig. 1, middle), which confirmed the previous result by Buday and Downward (25). These results indicate that Ras-GRF promotes the guanine nucleotide exchange reaction of R-Ras. Recently, we showed that C3G, a Crk SH3 domain-binding GRF, having a sequence similarity to mSos and Ras-GRF, is a GRF for Rap1 (15). Then we examined the guanine nucleotide exchange activity of C3G for R-Ras. As previously reported, C3G effectively activated Rap1B while it slightly activated Ha-Ras and N-Ras. When R-Ras was used as a substrate, the activity was intermediate between Ras and Rap1 (Fig. 1, right). This result shows that C3G also has a guanine nucleotide exchange activity for R-Ras.
The time courses of GDP release from R-Ras and Ha-Ras were measured in the presence or absence of Ras-GRF (Fig. 2). The spontaneous dissociation of GDP from R-Ras proceeded rather faster than that from Ha-Ras. Ras-GRF accelerated the GDP release from both small GTPases.
Ras-GRF and C3G Activate R-Ras in Intact Cells-We next examined whether Ras-GRF and C3G activate R-Ras in intact COS7 cells. A mammalian GST-fusion R-Ras expression vector was introduced into COS7 cells and guanine nucleotides bound to R-Ras were examined (Fig. 3, A and B). When R-Ras and a control vector pCAGGS were introduced, the relative amount of GTP-bound form was 10% of the total R-Ras. We then examined the effects of mSos, Ras-GRF, and C3G with or without the farnesylation signal of c-Ki-ras2 (CAAX box) on the activation of R-Ras. When R-Ras was expressed with Ras-GRF or Ras-GRF-F (with a farnesylation signal), the ratio of the GTPbound form of R-Ras greatly increased. Ras-GRF and Ras-GRF-F showed similar effects in the accumulation of the GTPbound form of R-Ras. C3G and C3G-F also increased the GTPbound form of R-Ras. Under the same conditions, mSos and mSos-F marginally increased the proportion of the GTP-bound form of R-Ras.
The effects of Ras-GRF, mSos, and C3G on the activation of Ha-Ras in COS7 cells were also examined. The basal level of the GTP-bound form was less than 5% of the total Ras. In this analysis, the Ras-GRF and mSos activation of Ras was stronger than that of C3G (Fig. 4).
These results demonstrate that Ras-GRF and C3G activate R-Ras as well as Ras in intact COS7 cells, while mSos activates Ha-Ras but not R-Ras.
Enhancement of Ras-GRF Activity by Calcium-Recently Farnsworth et al. (24) showed that the ability of Ras-GRF to activate Ras in vivo is markedly enhanced by a raised Ca 2ϩ concentration. Hence we next investigated whether the activation of R-Ras by Ras-GRF was also enhanced by the increase in the intracellular concentration of calcium (Fig. 5). When ionomycin was added to COS7 cells expressing GST-R-Ras and Ras-GRF, the GTP-bound form of R-Ras further increased. This result indicated that the activity of Ras-GRF toward R-Ras is also up-regulated by intracellular Ca 2ϩ concentrations.

R-Ras Is Expressed in Various
Organs-Previous study showed that Ras-GRF is highly expressed in the brain (26). Our results implicate that Ras-GRF and C3G can regulate R-Ras. Then we analyzed the expression level of R-Ras, Ras-GRF, and mSos in each organ by using the reverse transcribed PCR method. As shown in Fig. 6, Ras-GRF was expressed abundantly in brain and moderately in testis. In contrast to the restricted expression of Ras-GRF, R-Ras and mSos were ubiquitously expressed in all the organs examined. We previously showed that C3G is also ubiquitously expressed in all organs (14). These results suggest that Ras-GRF and C3G may act as GRFs for R-Ras in brain and testis, and in many tissues, respectively.

DISCUSSION
Genetic and biochemical studies established that CDC25 and Sos, and their counterparts in mammalian cells, are GRFs for Ras proteins (reviewed in Ref. 3). For Ral and Rap1, RalGDS (16) and C3G (15) have been identified as GRFs, respectively. However, a GRF for R-Ras has not yet been described. In the present study, we demonstrated that both Ras-GRF and C3G can activate R-Ras.
While Ras-GRF is expressed only in brain and testis, R-Ras and C3G are widely expressed in all the tissues examined ( Fig.  6) (9,14,26). Recently, GRF2, the amino acid sequence of which is 80% identical to that of Ras-GRF, was identified (27). GRF2 is widely expressed in various tissues and has GRF activity for Ras (27). Considering the high sequence similarity between Ras-GRF and GRF2, it is highly likely that GRF2 also acts on R-Ras as a GRF. Thus it is conceivable that R-Ras may be activated by Ras-GRF in brain and testis, and by C3G and GRF2 in various tissues. However, this does not exclude the presence of R-Ras-specific GRFs yet to be discovered.
Huff et al. (28) described that the coexpression of known Ras activators with R-Ras did not stimulate the transforming and the transactivating activities of R-Ras. Based on these results they argued that mSos and Ras-GRF are not the activators of R-Ras. However, these activities of R-Ras, even activated by a point mutation analogous to Ras val12 , are much weaker than those of activated Ras (8). When the wild type R-Ras and Ras-GRF are coexpressed, the relative amount of GTP-bound form of R-Ras was about 30% (Fig. 3), the value of which may be lower than that expected for the activated R-Ras. Therefore it may be difficult to determine by these rather indirect methods whether the activators of Ras also activate R-Ras. In contrast we demonstrated the activation of R-Ras by directly measuring the ratio of GTP-bound form.
The amino acid sequences of the catalytic domains of mSos, Ras-GRF, and C3G show about 30% identity with one another (14). However, there is a marked difference in their substrate specificities. Both mSos and Ras-GRF activated Ras efficiently. However, while Ras-GRF activated R-Ras as thoroughly as Ras, mSos did not activate R-Ras at all. C3G activated most efficiently Rap1, moderately R-Ras, and rather weakly Ras (this study) (14,15).
Residues of Ha-Ras and RAS2 that are critical for the interaction with GRFs have been identified using yeast (29 -32) and mammalian (33)(34)(35)  depending on the bound guanine nucleotides, another portion of Ras is obviously necessary for the interaction. Among 17 residues in the switch II region, 12 residues are identical between Ras and R-Ras, whereas 8 residues are conserved between R-Ras and Rap1. This may be the reason why Ras-GRF activates both Ras and R-Ras but not Rap1. Quilliam et al. (35) demonstrated that mutations in this region impair the sensitivities of the mutated Ras proteins to both Ras-GRF and mSos to similar extents, suggesting that Ras-GRF and mSos interact with the similar structure of Ras. However, Ras-GRF but not mSos activates R-Ras, demonstrating that there may be still some difference in the structures recognized by the two factors. Nine residues are identical between switch II regions of Ras and Rap1. Although critical residues of Rap1, R-Ras, and Ras for the interaction with C3G are yet to be identified, C3G showed the broadest substrate specificity.
Ras-GRF with or without a membrane-targeting signal exhibited similar activities in the stimulation of both Ha-Ras and R-Ras, whereas mSos with the targeting signal showed higher activity than that without the signal in the activation of Ha- Ras (Fig. 4). The difference in the effects of the membrane targeting signal may reflect the difference in the cellular localizations of Ras-GRF and mSos. It has been shown that mSos translocates from the cytosol to the plasma membrane by the adapter proteins, Grb2 and Shc, which causes the activation of Ras (3). Buchsbaum et al. (37) stated that Ras-GRF expressed in 293T cells distributes mainly in the particulate fraction, although a close relative of Ras-GRF, GRF2, expressed in the same cells translocates from the cytosol to the membrane periphery when the cells are treated with ionomycin (27). Activity of Ras-GRF is up-regulated by serum, lysophosphatidic acid, and calcium (24,38,39). Thus the activities of these Ras activators may be regulated by different mechanisms.
The fact that both Ras-GRF and C3G activate two distinct GTP-binding proteins is not surprising considering the high sequence resemblance among Ras family proteins. All the known Ras GAPs, p120GAP, neurofibromin, Gap1 m , and R-Ras GAP activate the GTPase activities of both Ras and R-Ras (7,17). 2 Despite the close resemblance to Ras, activated Rap1 inhibits GAP activity of p120GAP (40). Although not shown, if activated Rap1 also inhibits other Ras GAP molecules and R-Ras GAP, the activation of Rap1 by C3G would favor the formation of a GTP-bound form of R-Ras. Since the in vitro activation of R-Ras by C3G is rather modest compared with the activation of R-Ras in COS7 cells, this indirect mechanism might work in the cells.
It is very intriguing that Ras-GRF and C3G activate R-Ras and Ras, and Rap1 and R-Ras, respectively, since a single stimulus may activate two different signaling pathways. R-Ras promotes apoptotic cell death when interleukin-3 is withdrawn (12), whereas the expression of activated form of Ras prevents apoptotic cell death caused by interleukin-3 deprivation (41). In this case Ras and R-Ras cause opposite cell biological effects. In contrast, the activated forms of both R-Ras and Ras induce transformation of a particular fibroblast cell line (8,9). Hence the biological meaning of co-activation of R-Ras and Ras should await further experiments. Similarly, C3G activates Rap1 and R-Ras. Although Rap1 acts antagonistically against Ras function (42), it is not well investigated whether Rap1 and R-Ras interfere with each other in some biological processes.
Recently, it has become a consensus that the activation of low molecular weight GTP-binding proteins is achieved through the functional activation of GRFs rather than the inhibition of GTPase-activating proteins. For Ras, it has been demonstrated that the signals from receptor-type tyrosine kinases and cytosolic tyrosine kinases are transduced to Ras through the translocation of Sos (3). The activity of Ras-GRF is up-regulated by serum or lysophosphatidic acid (38,39) and by the increase in the intracellular calcium concentration (24). It was also demonstrated that a signal from the ␤␥ subunit of trimeric G proteins activates Ras-GRF in a phosphorylationdependent manner (43). Activation of Rap1 may be under the control of tyrosine kinases, since its GRF, C3G, is a Crk-binding protein.
Although not yet proven, these signals may also activate R-Ras. FIG. 5. Enhancement of Ras-GRF activity to R-Ras by calcium in COS7 cells. An expression vector for R-Ras was transfected into COS7 cells with either a control vector (pCAGGS) or an expression vector for Ras-GRF or Ras-GRF-F. After 48 h, cells were treated with or without 10 M ionomycin for 5 min, then guanine nucleotides bound to R-Ras were analyzed as in the legend to Fig. 3. Mean values obtained from three independent experiments were shown with standard deviations.
FIG. 6. The tissue specificities of R-Ras, Ras-GRF, and mSos mRNA form various organs were reverse transcribed and then amplified by PCR as described under "Experimental Procedures." The products were analyzed on 1.2% agarose gels, and the gels were stained with SYBER Green I (Molecular Probes).