A New Function of p120-GTPase-activating Protein

This work studies the coordination of the action of GTPase-activating protein (GAP) and guanine nucleotide exchange factor (GEF) on activated human c-Ha-Ras p21. Purified human p120-GAP was obtained with a new efficient procedure. To distinguish the GTPase-activating effect of p120-GAP from other effects dependent on the interaction with activated Ha-Ras, the nonhydrolyzable GTP analogue guanosine 5′-O-(thiotriphosphate) (GTPγS) was used. The results showed that the GTPγS/GTPγS exchange enhanced by the C-terminal catalytic domain of the yeast GEF Sdc25p (C-Sdc25p) is prevented by p120-GAP. This effect is strictly specific for the activated form of Ha-Ras, the target of GAP; no effect on Ha-Ras·GDP was detectable. The GAP catalytic domain also inhibited C-Sdc25p but to a lower extent. The interfering effect by p120-GAP was also evident in a homologous mammalian system, using full-length mouse RasGEF, its C-terminal half-molecule, or C-terminal catalytic domain. As a consequence of this inhibition, presence of p120-GAP enhanced the regeneration of Ha-Ras·GTPγS by GEF at a GDP:GTPγS ratio mimicking the in vivo GDP:GTP ratio. Our work describes a novel function of p120-GAP and suggests a mechanism by which GAP protects Ha-Ras·GTP in vivo against unproductive exchanges. This constrain is likely involved in the regulation of the physiological GDP/GTP cycle of Ras and in the action of p120-GAP as downstream effector of Ras. Helix α3 is proposed as a Ras element playing a key-role in the interference between GAP and GEF on Ras.

Ras proteins are molecular switches of a pathway regulating cell growth and differentiation by cycling between two conformations: the active GTP-bound state and the inactive GDPbound state (1). As for most GTPases, the GDP/GTP cycle of Ha-Ras is controlled by two kinds of regulators, the GTPase activating protein (GAP) 1 and the GDP/GTP exchange factor (GEF). Thus, the level of Ha-Ras⅐GTP depends on the ratio of the activities of these two regulators and consequently, conditions influencing their activity affect the function of Ha-Ras. The mechanisms involved in the regulation of these two ligands have only been in part clarified. For instance, why only in particular conditions is the decrease in GAP activity associated with the increase in the concentration of p21⅐GTP (2,3)? Why can overexpression of Ras be associated with a very low percentage of activated GTP-bound state (Ͻ0.3%) (4)? Upon activation of receptor or nonreceptor tyrosine kinase, p120-GAP is phosphorylated in vivo, but it is as yet unclear whether this modification is involved in the regulation of its negative effect on Ha-Ras⅐GTP (5). Concerning GEF, the activation of Ha-Ras by tyrosine kinase-linked receptors has been reported to depend on the transient translocation of the ubiquitous Ras⅐GEF SOS to the cell membrane (5), a process whose regulation still presents several unclear aspects. Recent work in vivo has suggested that Ca 2ϩ , calmodulin (6), and phosphorylation (7) control the activity of the neuronal CDC25-like RasGEF, but how these effects are coordinated, remains an open question. The biochemical characterization of CDC25 Mm has shown that the activity of its C-terminal catalytic domain is negatively regulated by the N-terminal moiety and that the Ca 2ϩ -dependent proteins calmodulin and calpain may be involved in this mechanism (8). Another important aspect, the effect of the simultaneous action of p120-GAP and GEF on the activated Ha-Ras state p21⅐GTP, has yet to be investigated. Indeed, GAP and GEF have both as common target GTP-bound Ha-Ras, with the specific difference that p120-GAP has a much higher affinity for Ha-Ras⅐GTP than for Ha-Ras⅐GDP (9), whereas GEF acts on Ras⅐GTP and Ras⅐GDP with nearly the same efficiency (10 -13). In this context, it is also worth mentioning that the GTP-bound elongation factor (EF) Tu, a model GTPase in bacterial protein biosynthesis sharing structural and functional homology with Ras, is protected against the action of its GEF EF-Ts by amino acid tRNA and the ribosome that is since long known to exert a GAP-like activity on the EF-Tu GTPase (14,15). The importance of a coordination between the various Ras ligands is moreover suggested by the increasing number of ligands found to interact with Ha-Ras besides GAPs and GEFs (for references, see Wittinghofer and Hermann (16)).
Starting from these considerations, we have studied the influence of GAP on the action of GEF on activated Ha-Ras. For this purpose, we have tried to distinguish possible effects of GAP from its ability to enhance the Ha-Ras-dependent GTPase by substituting GTP with its nonhydrolyzable analogue GTP␥S. As a system, we have chosen human p120-GAP, prepared by a novel method, and the catalytic domain of Saccharomyces cerevisiae Sdc25p (C-Sdc25p), a model GEF acting equally well on Ha-Ras p21, and Ras2p (10,12,17,18) or the mammalian RasGEF CDC25 Mm (p140-GRF). The results suggest that p120-GAP exerts an additional function that prevents GEF from interacting with p21⅐GTP. The evidence that this effect is involved in the modulation of the signals of the Ras pathway is discussed.

MATERIALS AND METHODS
Expression and Purification of Full-length p120-GAP-The Escherichia coli strain SCS-1 was transformed with a pGEX vector containing the NaeI-EcoRI fragment from pUC101, a vector kindly provided by Dr. F. McCormick. It encoded a p120-GAP lacking the first 5 amino acids, thus comprising residues 6 -1047. The bacteria were grown at 37°C in 800 ml of LB medium containing 50 g/ml ampicillin. Expression of the GST-fused protein was induced overnight at 24°C by adding low amounts of IPTG (0.01 mM) at a cell density of 0.4 A 600 unit. The cell culture, chilled to 4°C at 2.0 A 600 , was centrifugated at 4,000 ϫ g for 10 min, and the pellet was resuspended in 70 ml of buffer A (150 mM NaCl, 16 mM Na 2 HPO 4 , 4 mM NaH 2 PO 4 , pH 7.3) containing 2 mM phenylmethylsulfonyl fluoride. After sonication and addition of 1% triton, the suspension was centrifuged at 17,000 ϫ g for 30 min and the supernatant was mixed batchwise with 4 ml of glutathione-Sepharose 4B (Pharmacia), gently shaken for 1 h at 4°C. The resin was washed twice with buffer A and the protein eluted with glutathione (3 mg/ml) was concentrated to 1.5 ml with Aquacide II (Calbiochem) and then to 400 l on a Centricon-100 (Amicon) at 3,000 ϫ g, diluting several times with 50 mM Tris-HCl, pH 7.5, 50 mM NaCl and 28 mM 2-mercaptoethanol. The protein solution, mixed with an identical volume of glycerol, was stable for at least several months when stored at Ϫ20°C.
The GST fusion was cleaved by incubation with 0.1 unit of thrombin (Sigma), 1 g of protein in 25 mM Tris-HCl, pH 8.0, 25 mM NaCl, 25 mM CaCl 2 , 7 mM mercaptoethanol, and 10% glycerol at 30°C for 30 min. The thrombin was removed by washing on Centricon-100 as above.
Dissociation Rate Constants-Dissociation of the p21⅐nucleotide complexes was determined at 30°C by the nitrocellulose binding assay. The p21⅐[ 3 H]GDP and p21⅐[␥-35 S]GTP␥S complexes were obtained by incubation of Ha-Ras p21 with a 4-fold excess of [ 3 H]GDP (340 Bq⅐pmol Ϫ1 ) or [␥-35 S]GTP␥S (44.8 Bq⅐pmol Ϫ1 ), respectively, for 10 min at 30°C in buffer B (50 mM Tris-HCl, pH 7.5, 1 mM MgCl 2 , 3 mM EDTA, 1 mM dithiothreitol and 1 mg/ml bovine serum albumin). MgCl 2 (final concentration, 3 mM) was then added to stabilize the p21⅐nucleotide complex. The dissociation was started by addition of a 500-fold excess of cold nucleotide with or without GEF as indicated in the legends to figures. At time intervals, aliquots (10 l) were withdrawn, passed on nitrocellulose filters (0.45 m of Sartorius SM 11 306), and the retained radioactivity measured in a liquid scintillation counter (Pharmacia, model Wallac 1410). The rate constants were calculated as described (13,19).
For competition assays between C-Sdc25p or CDC25 Mm and p120-GAP, the dissociation rate of p21⅐[ 3 H]GDP or p21⅐[␥-35 S]GTP␥S, preincubated or not with various concentrations of p120-GAP, was measured in the absence and in the presence of different concentrations of GEF. For details see legends to figures.
Specific Activity of p120-GAP-The p120-GAP activity was measured at a concentration of 10 nM Ha-Ras p21-bound GTP and the obtained values extrapolated to 1 M Ha-Ras p21-bound GTP for comparison with the data of the literature. One unit of specific activity of Ha-Ras p21 stimulation by GAP is defined as the amount of GAP inducing the hydrolysis of 1 nmol of Ha-Ras p21⅐GTP/min at 25°C (20).
Other Methods and Materials-The catalytic domain of p120-GAP (C-terminal 350 amino acid residues) was purified as GST fusion protein with the same method as p120-GAP. C-Sdc25p (C-terminal 550 amino acid residues) was produced and purified to near homogeneity as described previously (12). Purified Ha-Ras p21 (21), full-length CDC25 Mm (1262 amino acid residues), its C-terminal half-molecule of 631 residues (C-CDC25 Mm631 ) (8) and the catalytic domain of 285 residues (C-CDC25 Mm285 ) (13) were obtained as reported. SDS-PAGE was carried out using 7.5% acrylamide gel (22). Immunoblot analysis was performed using specific antibodies anti-GST produced in rabbit and peroxidase-conjugated antibodies revealed by the diaminobezamidine method. Protein concentration was determined by the Bio-Rad protein assay, using BSA as standard, and in the case of p21 also checked by GDP binding.

RESULTS
Overproduction and Purification of p120-GAP-The pGEX system allowed an easy reproducible production and purification of the recombinant human p120-GAP in good yield and, noteworthy, in the absence of proteolytic phenomena (Fig. 1A). As shown by Western blot analysis with specific antibodies anti-GST (Fig. 1B), virtually 90% of the produced GST-p120-GAP remained in the supernatant fraction, after centrifugation for 30 min at 17,000 ϫ g. The temperature of induction and the concentration of IPTG were important parameters to obtain high overproduction and solubility. In fact, induction at 24°C and 0.01 mM IPTG yielded higher production and solubility of GST-p120-GAP than at 37°C and 0.1 mM IPTG (not shown). No partial translation products of p120-GAP were detectable. After the affinity chromatography and Centricon-100 treatment, the GST-p120-GAP was at least 90% pure and the yield was ϳ1 mg/L for cell culture of 2.0 A 600 density (Fig. 1A, lane 5). The contamination essentially consisted of a faster migrating component unreactive to GST antibodies. Purified GST-p120-GAP was stable for several months, when kept at Ϫ20°C in storage buffer.
Stimulation by C-Sdc25p of the Ha-Ras GTP␥S/GTP␥S Exchange Is Specifically Inhibited by p120-GAP and by Its Catalytic Domain C-GAP-To determine the effect of GAP on the activity of the exchange factor on both active and inactive states of Ras, the complex formed by p21 with the nonhydrolyzable GTP analogue [␥-35 S]GTP␥S or with [ 3 H]GDP was used. It is known that p120-GAP and C-GAP have an affinity for GTP-bound p21 much higher (Ͼ100 times) than for GDPbound p21 (9,24).  (Fig. 2B). The inhibition by p120-GAP of the action of C-Sdc25p increased with increasing the concentration of p120-GAP (Figs. 3, A and B). A similar inhibitory effect by p120-GAP was also observed on the the stimulation by C-Sdc25p of the GTP␥S/GDP exchange reaction of Ha-Ras (not illustrated). We could estimate that the IC 50 of p120-GAP on the C-Sdc25p activity was around 140 nM (Fig. 3B). When we replaced p120-GAP with its isolated catalytic domain C-GAP (350 amino acid), the inhibition of the C-Sdc25p was about 2.5-fold lower (Fig. 3, C and D), the IC 50 being approximately 350 nM. This lower effect is in good agreement with the 2-fold lower times affinity observed for a similar catalytic domain of p120-GAP, GAP334, as reported previously (20).
To our surprise, in the course of these measurements we observed a slight but reproducible stimulation of the intrinsic GTP␥S/GTP␥S exchange of Ha-Ras by p120-GAP, an effect that increased with increasing the concentration of p120-GAP and was not observed if C-GAP replaced p120-GAP (cf. Fig. 3, A and  C). Worth mentioning, no p120-GAP-dependent hydrolysis of [␥-35 S]GTP␥S could be detected (not shown).
In the set of experiments described of Fig. 4, we examined the dependence of the GEF activity on the molar ratio C-Sdc25p:p120-GAP, using increasing concentrations of C-Sdc25p and a fixed amount of p120-GAP (136 nM), corresponding to the concentration giving approximately a 50% inhibition in the experiments of Fig. 3. An increase in C-Sdc25p from 10 to 100 nM considerably reduced the inhibition by p120-GAP, as shown by the decrease in the difference between the kЈ Ϫ1 values in the absence and in the presence of p120-GAP (Fig. 4, I-IV). In fact, with the lowest concentration of C-Sdc25p (10 nM) the inhibition by p120-GAP on the dissociation rate constant of p21⅐[␥-35 S]GTP␥S was nearly 70%, whereas in the presence of 100 nM C-Sdc25p the inhibition by p120-GAP was practically absent. p120-GAP also Inhibits the GEF Activity of Mammalian CDC25 Mm -The interference by GAP on the GEF activity was also tested in a homologous mammalian system by using three forms of the mammalian RasGEF mouse CDC25 Mm : the fulllength molecule, the C-terminal half-molecule, and the short catalytic domain CDC25 Mm285 . Noteworthy, the full-length CDC25 Mm has a GEF activity 5 times lower than CDC25 Mm631 and 25 times less than CDC25 Mm285 (8). Also in this system, p120-GAP exerted an inhibitory activity, in which specific differences depending on the GEF form could be observed. As shown in Fig. 5, in which the concentration of the diverse CDC25 Mm forms was chosen to give similar dissociation rates, the inhibition of the stimulation of the Ha-Ras GTP␥S/GTP␥S exchange in the case of full-length CDC25 Mm and C-terminal half-molecule was about 50%. The effect was lower (inhibition, 18%) when the short catalytic domain CDC25 Mm285 was used.
Presence of p120-GAP Enhances the Regeneration of the Ha-Ras Active Form by GEF-To evaluate the implication of the inhibitory effect of p120-GAP on GEF in the regeneration of the active form of Ha-Ras and, consequently, to examine the possibility of correlating these observations in vitro with the physiological conditions in the cell, we have investigated the influence of p120-GAP under conditions in which the molar ratio of the free nucleotides GDP and GTP (as GTP␥S) corresponded to the range of values estimated to occur in the cell. As shown in Fig. 6, at a GDP to GTP␥S ratio of 1 to 10, the addition of 130 nM p120-GAP enhanced the CDC25 Mm -dependent stimulation of the GDP/GTP␥S exchange on Ha-Ras. This result supports the possibility that the inhibition by GAP of the GEF activity via sequestration of the active form of Ha-Ras favors the interaction between the exchange factor and Ha-Ras⅐GDP, its physiological substrate. DISCUSSION The increasing number of interactions attributed to Ha-Ras, a central crossing point for the regulation of the signal transmission in the cell, suggests the existence of mechanisms coordinating the complex pattern of interactions of ligands on Ha-Ras (for references, see Wittinghofer and Hermann (16)). In an attempt to shed some light on possible mechanisms, we have chosen a system in vitro analyzing whether GAP can directly affect the activation of Ha-Ras by GEF, since GAP and GEF are both known to affect the GTP-bound Ha-Ras. For this study, it was important to have available the whole molecule of p120-GAP. We therefore developed a simple procedure to isolate highly purified p120-GAP, by defining precise, reproducible conditions for its efficient production in E. coli. Despite the lack of the first 5 residues, the ability of our p120-GAP to stimulate the p21 GTPase was virtually the same as reported for fulllength p120-GAP produced and purified from baculovirus/sf9 insect cells (20,23). No proteolytic phenomena were observed during its preparation and the stability of the purified form was verified over months. This method facilitates the utiliza- tion of this regulator of Ras, whose isolation is known to present still considerable difficulties (25), avoiding time-consuming and expensive procedures, such as the use of the insect cell system. The good yield of p120-GAP obtained with this method makes possible its use for future structural studies.
In the GDP/GTP cycle of Ras proteins, the reversibility of all the partial steps of the nucleotide⅐p21 interaction, except for the hydrolysis of GTP, could allow the occurrence of unproductive reactions, such as the GTP/GTP exchange. In fact, in the cell the only physiologically significant exchange is the replacement of the Ha-Ras-bound GDP with GTP. Constraints have been proposed for promoting the GDP/GTP exchange on Ha-Ras, such as the higher concentration of GTP versus that of GDP in the cell (11,26) and the somewhat higher affinity of GEF for the GDP-versus the GTP-bound state of Ha-Ras (10 -13). These properties, however, are not sufficient to guarantee alone a physiologically productive nucleotide exchange on Ha-Ras. In this work, we report an additional function of p120-GAP, that might have regulatory significance in the cell. The use of the nonhydrolyzable GTP analogue, GTP␥S, that abolishes the stimulatory effect of p120-GAP on Ha-Ras, allowed us to reveal a new effect associated with the active form of Ha-Ras, the privileged target of GAP; p120-GAP bound to p21⅐GTP␥S can prevent the interaction between GEF and the active form of Ha-Ras, and thus the stimulation of the GTP␥S/GTP␥S exchange. Also C-GAP inhibited the GEF activity to an extent 2.5-fold as low, in agreement with the lower (ϳ2-fold) affinity for activated Ha-Ras (20). No inhibition of the GEF-dependent GDP/GDP exchange on p21 was observed.
These data suggest a mechanism, by which p120-GAP in the cell would prevent the GEF interaction with Ha-Ras⅐GTP after accomplishment of the GDP/GTP exchange on Ras. Since the role of GEF is to activate as many as possible molecules of Ras, the interaction with already activated Ras proteins is physiologically useless, delaying the productive GDP/GTP cycle. The p120-GAP protection of Ha-Ras would increase the probability of GEF to interact with its biologically important substrate, the Ras⅐GDP complex. Indeed, in agreement with this model, under conditions mimicking the cellular GDP:GTP molar ratio of 1: 10, the presence of p120-GAP can stimulate the GEF-dependent GDP/GTP␥S exchange on Ha-Ras. As enhancer of the GTPase activity of Ras promoting the return to the inactive GDP-bound state, p120-GAP has been considered a negative regulator of Ras. However, inhibitory effects on the ability of this GAP to enhance the GTPase activity of Ha-Ras have been reported as response to extracellular stimuli. For example, stimulation of the T cell receptors in T lymphocytes (2) or treatment with erythropoietin of human erythroleukemia cells (3) results in a large increase in Ras⅐GTP as well as in a decrease in the stimulation of the GTPase activity by GAP in cell lysate. p120-GAP is phosphorylated in vivo upon activation of receptors or nonreceptors tyrosine kinases. Phosphorylated p120-GAP associated with the p190 phosphoprotein was found to have a decreased GTPase activating effect (27). Mitogenic lipids such as phosphatidic acid and arachidonic acid can inhibit p120-GAP in vitro (28). All these data are consistent with the possibility that under specific in vivo conditions the GTPase activating effect of p120-GAP may be inhibited. Evidence also exists that p120-GAP may act as downstream effector (for refs, cf. Pronk and Bos (5). The dual role of p120-GAP as negative regulator and effector of Ras is far from being fully understood and it is possible that the protective association of p120-GAP with activated p21 is correlated with the function of GAP as transmitter of downstream signals of Ha-Ras.
As model GEF we have used for most experiments the highly purified catalytic domain of S. cerevisiae Sdc25p that is the best biochemically characterized exchange factor so far, but we confirmed the competition between GAP and GEF on the active form of Ha-Ras in a homologous mammalian system using the mouse GEF CDC25 Mm . The GEF activity of full-length CDC25 Mm and of the C-terminal half-molecule, CDC25 Mm631 , were inhibited by p120-GAP more efficiently than C-Sdc25p, in agreement with their lower affinity for Ha-Ras. The observation that the inhibitory effect on the isolated catalytic domain CDC25 Mm285 is weak suggests that noncatalytic domains of GEF are also involved in the interference with GAP.
The observed reproducible slight stimulation by p120-GAP of the intrinsic GTP␥S exchange of Ha-Ras was not induced by C-GAP. Since the noncatalytic regions of p120-GAP are needed for a fully productive Ras-GAP interaction (20), it is possible that these regions may induce some destabilization of the Rasnucleotide interaction, very likely via intramolecular arrangement of the Ras nucleotide binding pocket.
We have demonstrated the inhibition by GAP of the GEF activity in vitro. The reciprocal effect, the influence of GEF on the GAP activity is also likely to exist, as suggested by preliminary experiments indicating an inhibition by C-Sdc25p of the p120-GAP-dependent stimulation of the Ha-Ras p21⅐GTPase. 2 Since the investigation of this aspect presents several technical difficulties, because the GTP hydrolysis produces the inactive form p21⅐GDP that is also a substrate of the competitor GEF, other experimental approaches may be necessary to define in an unequivocal manner the reciprocal nature of the constraints induced by GEF on GAP.
Since it is known that besides GEF and GAP, the active form of Ha-Ras can bind other ligands, such as c-Raf, Rin, protein kinase C, and P1(3)K (for references, see Wittinghofer and Hermann (16)), it is likely that not only GAP but also these effectors contribute to protect p21⅐GTP, hindering futile reactions by favoring a physiological course of the GEF-dependent GDP/GTP exchange reaction. In the cell the action of these effector molecules could equal that of p120-GAP or be even more relevant. In this regard, it is worth mentioning that the Ras-binding domain of c-Raf was found to compete with p120-GAP for binding to Ha-Ras⅐GTP (29 -31). The coordination of these multiple interactions could represent an important mechanism for the regulation of the level of the active and inactive state of Ha-Ras.
What is the structural background of the interference of GAP on the GEF binding to Ha-Ras? Does this effect depend: on (i) the overlapping of GAP and GEF binding sites on p21 inducing hindrance phenomena, or on (ii) allosteric long range effects induced by GAP on the GEF binding site? The specificity of the p120-GAP effect suggests that this competition is mediated by the regions of p21 mostly influenced by the active and the inactive states of Ras proteins, namely the switch I (loop L2/ N-terminal ␤2) and switch II regions (loop L4/helix ␣2). Among the regions of Ras known so far to be essential for the function of GAP, the loop L2 plus the N-terminal strand ␤2 and the loop L6/N-terminal helix ␣3 are structural elements implicated in the binding site of GAP (32)(33)(34)(35). Loop L1 has been shown to play a key role in the molecular mechanism of the GTPase activation (20, 24, 32, 36 -38). As this loop contains a residue (Ala 11 in Ha-Ras and Gly 18 in Ras2p) important for the specificity toward RasGAP, it can be considered as potential binding site (35). Concerning the interaction between Ras and GEF, a direct role has been reported for helix ␣2 (12, 39 -45) and the region spanning residues 101-105, that corresponds to the C-terminal helix ␣3/N-terminal loop L7 (46,47). Therefore, the present knowledge of the Ras interaction sites with GAP and GEF indicates that helix ␣3 represents a secondary element concerning both regulators, GAP interacting with its N-termi-nal part and GEF with the C-terminal one (Fig. 7). The observation that helix ␣3 play an important role in the binding of GAP and GEF could explain the interference between these two kinds of regulators. It is also important to mention that helix ␣3/loop L7 has been shown by three-dimensional studies to be strictly coordinated with helix ␣2/loop L5 that on its turn is closely associated to the phosphoryl binding loop L2, an important element of the nucleotide binding pocket (48). In the tertiary model of Ras, the regions comprising helix ␣2/loop L5 and helix ␣3/loop L7 are located on the same surface side of the molecule.
In conclusion, the results of this work have enlightened a coordination mechanism between the exchange activity of GEF and a function of p120-GAP distinct from its action as enhancer of the p21 GTPase. This mechanism very probably contributes to determine the physiological course of the GDP/GTP cycle of Ras proteins and thus the function of the Ras pathway in the cell. We have reproduced in a system in vitro some effects that could take place in vivo, even though our system represents a simplified version of possible in vivo situations. The extension of this kind of study to the other ligands of Ras is a prerequisite for building up in vitro model systems capable of mimicking more faithfully the complexity of the regulatory mechanisms of Ha-Ras activity in the cell.