Distal switch II region of Ras2p is required for interaction with guanine nucleotide exchange factor.

The interaction of Saccharomyces cerevisiae Ras2p with the catalytic domain of the GDP/GTP exchange factors (GEFs) mouse CDC25Mm, yeast Cdc25p, and Sdc25p was analyzed by introducing the substitution R80D/N81D into Ras2p S24N, a mutant that is shown to interfere with the Ras2p wild type (wt)-GEF interaction by forming a stable complex. The triple mutant, like Ras2p R80D/N81D, did not interfere with the action of GEF on Ras2p wt (or H-Ras p21) and was unable to form a stable complex with GEF. The GEF stimulation of the nucleotide dissociation of the triple mutant was virtually abolished and strongly decreased with the double mutant. The affinity of Ras2p S24N/R80D/N81D for GDP and GTP was decreased 3 and 4 orders of magnitude, respectively, like that of Ras2p S24N, whereas the double mutant behaved as Ras2p wt. Like Ras2p S24N and unlike Ras2p R80D/N81D, the GTP-bound triple mutant did not activate adenylyl cyclase. Thus, the triple mutant and Ras2p S24N have opposite properties toward the binding to GEF but similarly modified behaviors toward GDP, GTP, and adenylyl cyclase. This work emphasizes the determinant role of the distal switch II region of Ras2p for the interaction with GEF and the different structural background of the interaction with adenylyl cyclase.

Ras proteins are GTPases that regulate cell growth and differentiation by cycling between the active GTP-bound and the inactive GDP-bound states. The level of these two forms is determined by GTPase-activating proteins and guanine nucleotide exchange factors (GEFs) 1 (Boguski & McCormick, 1993). The yeast Saccharomyces cerevisiae Cdc25p  was the prototype for Ras GEFs, such as S. cerevisiae Sdc25p (Créchet et al., 1990b and mammalian tissuesspecific brain GEFs Jacquet et al., 1992;Schweighoffer et al., 1993). Due to difficulties in isolating the intact protein, only the catalytic domain of these GEFs has been used to investigate the mechanism of the exchange reaction in vitro so far (Créchet et al., 1990b;Lai et al., 1993;Shou et al., 1992;Schweighoffer et al., 1993;Haney & Broach, 1994;Jacquet et al., 1994Jacquet et al., , 1995Poullet et al., 1995). Several reports have dealt in these past years with the structures of Ras proteins involved in the interaction with GEF. Although no definite conclusions could be drawn concerning the location of the binding site of Ras proteins for GEF, site-directed mutagenesis of Ras proteins has given useful information on the role played by the various regions of the Ras molecule in the action of GEF. Specific substitutions in the phosphoryl binding I region, the switch I and II regions of human c-H-Ras p21 and Ras2p were identified that influence negatively the action of GEF; some of them have little effect on the affinity for GEF (Mistou et al., 1992), whereas others can induce dominant negative interferences on the interaction between wild type Ras and GEF, due to the formation of a stable complex between the mutant and GEF (Powers et al., 1989;Szeberényi et al., 1990;Stacey et al., 1991;Hwang et al., 1993;Chen et al., 1994;Jung et al., 1994). Effects of a recessive type, which inhibit the response to and the formation of a stable complex with GEF (Verrotti et al., 1992;Poullet et al., 1995), were first observed with mutations in the distal part of helix ␣2 (residues 80 -82 in Ras2p) but negative effects, probably of related type, were also obtained by substituting other surface-located residues on helix ␣2 in p21 (Howe & Marshall, 1993;Mosteller et al., 1994;Quilliam et al., 1994) and in the boundary between helix ␣3 and loop L7 (Segal et al., 1995).
In S. cerevisiae, adenylyl cyclase is a major target of the GTP-bound form of Ras proteins and constitutes the first element of a cascade of kinases influencing the activity of transcription factors. Experimental evidence obtained from the yeast system indicates that substitutions in the distal switch II region induce little, if any, effect on the activation of adenylyl cyclase (Verrotti et al., 1992;Mirisola et al., 1994), whereas other observations suggest that modification of the helix ␣2 affect negatively the adenylyl cyclase activity (Segal et al., 1995). To examine the functional role of this region concerning the response to GEF, the mechanism of interference of dominant negative mutations, and the activation of adenylyl cyclase, we have constructed a Ras2p mutant that combines substitutions of dominant (S24N) and recessive type (R80D/ N81D). The biochemical analysis of this mutant, as compared with Ras2p wt, Ras2p S24N, and Ras2p R80D/N81D, 2 emphasizes the importance of the conserved distal switch II region of Ras2p for the action of GEF action and the specific properties of the mechanism of interference of Ras2p mutants.

MATERIALS AND METHODS
Preparation of Ras Proteins-Ras2p wt and mutants were produced in Escherichia coli strain SCS1 as recombinant protein fused with glutathione S-tranferase using the pGEX2T vector. Ras2p S24N was constructed as reported (Poullet et al., 1995). Ras2p R80D/N81D was obtained by cloning the NdeI-SalI fragment from pAVD80D81 (Verrotti et al., 1992) into the SmaI site of pGEX2T (pGEX2T R80D/N81D). Ras2p S24N/R80D/N81D was engineered by inserting the NdeI-PstI fragment of pAV1 carrying the S24N mutation (Poullet et al., 1995) as replacement for the homologous wild type segment in pGEX2T R80D/ N81D. The NdeI-SalI fragment from pSKc-Hras (Gross et al., 1985), containing the open reading frame of human c-H-ras p21 was cloned into the SmaI site of pGEX2T to express the glutathione S-transferase-H-Ras p21 fusion. The transformed E. coli strains were grown at 28°C in 2 liters of LB-rich medium containing 50 g/ml ampicillin. Cell cultures were induced at a density of 0.3 A 600 units with 0.1 mM isopropyl-␤-D-thiogalactopyranoside. After 12-15 h of growth, the cells were collected by centrifugation, washed, and sonicated thrice at 4°C, at which temperature all the subsequent steps were carried out (Poullet et al., 1995). The affinity chromatography on glutathione-agarose and the thrombin treatment to remove the fused glutathione S-transferase were carried out as reported (Poullet et al., 1995). The faster the purification procedure, the higher was the content of the intact Ras2p products. The purified preparations contained ϳ65% of full-length form plus ϳ30 and ϳ5% of the 29-and 37-kDa proteolytic forms, respectively. Mono-Q HR 5/5 chromatography (fast protein liquid chromatography, Pharmacia Biotech Inc.) using a linear 20 -300 mM KCl gradient (50 ml) in 25 mM Tris-HCl, pH 7.8, 2 mM MgCl 2 , 7 mM 2-mercaptoethanol, and 10 M GDP allowed the separation of nearly homogeneous full-length Ras2p from the 29-kDa form. The p21 was purified on Mono-Q HR 5/5 using the same gradient. After concentration in an Amicon ultrafiltration apparatus (Diaflo membrane PM10), the Ras proteins were stored at Ϫ20°C in 25 mM Tris-HCl, pH 7.5, 2 mM MgCl 2 , 30 mM KCl, 50% glycerol, 1 mM DTT, and 10 M GDP.
Ras2p⅐Nucleotide Interaction-The dissociation constants (K d ) and the dissociation rate constants (k Ϫ1 ) of Ras-nucleotide complexes were determined by the nitrocellulose binding assay using 0.45-m Sartorius SM 11 306 filters (Créchet and Parmeggiani, 1986). All tests were performed at 30°C in 50 mM Tris-HCl, pH 7.5, 100 mM NH 4 Cl, 1 mM DTT, and 0.05 mg/ml bovine serum albumin (buffer A) containing MgCl 2 as indicated in the legends to figures. The preformed labeled Ras⅐[ 3 H]GDP complexes were prepared by incubating for 20 min at 30°C 3 M Ras2p⅐GDP in 25 mM Tris-HCl, pH 7.5, 10 mM EDTA, 1 mM DTT, and 0.05 mg/ml bovine serum albumin with 5 M [ 3 H]GDP (200 GBq/mmol, Du Pont NEN). MgCl 2 (10 mM) was then added. The dissociation rates of the Ras⅐[ 3 H]GDP complexes were determined in 165 l of buffer A containing 68 nM Ras⅐[ 3 H]GDP, MgCl 2 , and GEF, as indicated in the legends to figures, and a 1000-fold molar excess of unlabeled nucleotide. The reaction was started with preformed Ras⅐[ 3 H]GDP complex. At time intervals, 25-l aliquots were filtered on nitrocellulose discs that were washed twice with 3 ml of ice-cold 50 mM Tris-HCl, pH 7.5, 100 mM NH 4 Cl, 10 mM MgCl 2 , and 7 mM 2-mercaptoethanol. The KЈ d of Ras2p⅐GTP and Ras2p⅐GDP complexes were calculated by Scatchard plots according to the equation: where r is the average mol of GTP or GDP bound per mol of Ras2p and n is the number of binding sites. To eliminate traces of GDP from commercial GTP preparations, the concentrated solution of labeled or unlabeled GTP used was incubated for 10 min at 30°C in a 40-l solution containing 20 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 20 mM NH 4 Cl, 1 mM DTT, 20 mM phosphoenolpyruvate, and 30 g of pyruvate kinase (Boehringer Mannheim) for 40 mM nucleotide. Formation of Ras2p⅐CDC25-C Mm Complexes-200 l of buffer B (25 mM Tris-HCl, pH 7.8, 1 mM MgCl 2 , 150 mM NaCl, and 7 mM 2-mercaptoethanol) containing GDP-free Ras2p in the 29-kDa form (ϳ250 pmol) and CDC25-C Mm in a 2-fold molar excess over Ras2p with or without GDP or GTP were loaded on a AcA 44 Ultrogel column (57 ϫ 0.9 cm) equilibrated with buffer B at 4°C. The flow rate controlled by the fast protein liquid chromatography system was 0.04 ml/min, and 200-l fractions were collected 250 min after loading. Ras2p and CDC25-C Mm were detected by determining the bound [ 3 H]GDP and the [ 3 H]GDP release activity, respectively. In the former case, 20-l aliquots were incubated for 10 min at 30°C with 2 M [ 3 H] GDP (25-180 GBq/mmol) in a 50-l reaction mixture (25 mM Tris-HCl, pH 7.5, 2 mM EDTA, 20 mM (NH 4 ) 2 SO 4 and 1 mM DTT). In the latter case, the reaction was started by adding 3 pmol of preformed Ras2p [ 3 H]GDP in 30 l of buffer B to a 20-l aliquot, to which 3 l containing 9 nmol of cold GDP were added. The decrease in radioactivity was measured after 15 min at 30°C. Immunostained Western blot using 140-l lyophilized fractions was performed according to Poullet et al. (1995) with Ras2p-specific antibodies and CDC25-C Mm antibodies obtained from rabbit immunized against purified CDC25-C Mm .
Other Methods-The purity of proteins was estimated by Coomassie Blue staining on 12% SDS-polyacrylamide gel electrophoresis, by the Bradford method (Bradford, 1976), and for Ras proteins their concentration was estimated by their ability to bind [ 3 H]GDP in the presence of a saturating concentration of the labeled substrate (0.7-1 mol/mol of Ras as compared with the Bradford method).

RESULTS
Properties of Ras2p S24N-In mammalian cells, the mutation S17N induces a dominant negative phenotype (Stacey et al., 1991). This was also the case with the S. cerevisiae homologous substitution S24N, as was shown by genetic analysis of the RAS2 S24N phenotype. 3 To characterize the interfering properties of Ras2p S24N in vitro in a quantitative manner, we examined its effect on the dissociation rate of Ras2p wt⅐[ 3 H]GDP induced by Sdc25p-C ( 1D). In all cases the response of Ras wt to GEF was inhibited by Ras2p S24N in a concentration-dependent manner. Because the K m for the exchange reaction of the three GEFs are different (Jacquet et al., 1994), their concentrations (50 nM Sdc25p-C, 2.2 nM Cdc25p-C, or 70 nM CDC25 Mm -C) were chosen to give a dissociation rate 12-20 times higher than the intrinsic dissociation of the [ 3 H]GDP⅐Ras complex. In order to avoid any interference between the intrinsically fast GDP dissociation of Ras2p S24N (see below) and the Ras wt⅐[ 3 H]GDP dissociation rate dependent on GEF, increasing concentrations of the 42-kDa form of Ras2p S24N⅐GDP were first preincubated with the various GEFs. A 1000-fold excess of cold GDP was then added, and the reaction was started with Ras2p wt⅐[ 3 H]GDP. The IC 50 of Ras2p S24N (Fig.  1E) was similar for all three GEFs used (45 nM for Sdc25p-C, 70 nM for Cdc25p-C, and 80 nM for CDC25 Mm -C), showing that its affinity for these GEFs lies in the same range. The last 112 C-terminal residues of Ras2p were not necessary for the interaction with the GEFs, because the 42-and 29-kDa forms of Ras2p S24N induced the same interfering effect (Fig. 1F).
The R80D/N81D Substitutions Abolish the S24N-induced Interference on the Ras2p-GEF Interaction-Substitution R80D/N81D was reported to abolish the Ras2p response to Sdc25p without affecting the intrinsic GDP/GTP exchange (Verrotti et al., 1992). The introduction of these two substitutions in Ras2p S24N (Fig. 2) caused the loss of the ability to interfere with the Cdc25p-C-mediated Ras2p wt⅐[ 3 H]GDP dissociation rate even at concentrations as high as 1.5 M, whereas in the same conditions 0.5 M Ras2p S24N reduced 12 times the dissociation rate of Ras2p wt⅐[ 3 H]GDP. This shows that the nature of residues Arg 80 and Asn 81 is essential for the interfering properties of Ras2p S24N.
The ability of the various Ras2p species to form stable complexes with Cdc25p on gel filtration was also analyzed. For these experiments, we used the truncated form of Ras2p to improve the resolution of the eluted products. CDC25 Mm -C was preferred to Cdc25p-C or Sdc25p-C, because unlike the purified forms of these two yeast GEFs (Poullet et al., 1995), it does not display hydrophobic interactions with the acrylamide-agarose matrix, leading to a retention on filtration chromatography. CDC25 Mm -C was added in a 2 to 1 molar excess over Ras2p. In these conditions, as deduced from the ability of the eluted fractions to bind [ 3 H]GDP and stimulate the dissociation of Ras2p⅐[ 3 H]GDP, more than 80% of the GDP free-Ras2p wt was engaged in a stable complex with CDC25 Mm -C (Fig. 3A). In contrast to Ras2p wt, the triple mutant did not form a complex with CDC25 Mm -C, even in the absence of nucleotides (Fig. 3B), like Ras2p R80D/N81D (Fig. 3C). As was previously shown with Sdc25p-C (Poullet et al., 1995), the complex Ras2p S24N⅐CDC25 Mm (Fig. 3D) required for dissociation much higher concentrations of nucleotide (0.1 mM GDP (Fig. 3E) or GTP (Fig. 3F)) than Ras2p wt (1 M, not shown). The effect of the R80D/N81D to relieve the S24N effect stresses the importance of Arg 80 and Asn 81 for the binding to GEF. The existence of a 1:1 stoichiometry between Ras2p and GEF in the stable complex was proved by immunotransfer Western blot analysis on SDS-polyacrylamide gel electrophoresis (not shown).
The Action of GEF on the Nucleotide Interaction of the Various Ras2p Species-The analysis of the interaction between the various mutants and GTP or GDP was carried out in the presence of 10 mM MgCl 2 . The stabilization by this cation of the Ras-nucleotide complex (De Vendittis et al., 1986a;Hall & Self, 1986) was required to obtain dissociation rates measurable by the nitrocellulose binding assay in the case of high concentrations of GEF or fast intrinsic dissociation rates as found for Ras2p S24N (see below). Table I shows that the intrinsic dissociation rate of the GDP complex of the triple mutant was 300 times faster than that of the Ras2p wt complex, corresponding to the fast intrinsic dissociation of Ras2p S24N⅐GDP, whose rate could, however, still be enhanced by CDC25 Mm , to a 10-fold smaller extent than Ras2p wt, whereas the rate of the triple mutant was virtually insensitive to GEF.
Because even in the presence of 10 MgCl 2 the dissociation rates of the GTP complexes of the triple and S24N mutants were too fast to be measured by the nitrocellulose binding assay, the analysis of these complexes was limited to the calculation of the dissociation constants (KЈ d ) from Scatchard plots, obtained with nucleotide-free Ras2p and GTP concentrations varying between 4 and 80 M (Table II). The KЈ d values of the GTP complex of Ras2p S24N (14.4 M) and the triple mutant (97 M) lay in the same range and were much higher than that of Ras2p wt (3 nM, Créchet et al., 1990a). The KЈ d values of their GDP complexes, as determined for concentrations of GDP between 0.2 and 6 M, were close (1.6 M for Ras2p S24N and 2.5 M for the triple mutant) and 1 order of magnitude smaller than those of the GTP complexes.
In agreement with our previous observations (Verrotti et al., 1992), the double mutant displayed virtually the same intrinsic GDP "off" rate as Ras2p wt. As shown in Table I, its response to GEF was not totally abolished, unlike that of the triple mutant. A concentration of 0.83 M CDC25 Mm -C, stimulating 250 times the dissociation rate of Ras2p wt⅐GDP, was found to increase six times the dissociation rate of the GDP complex of the double mutant. Therefore, the strength of its interaction was 40 times weaker than that of Ras2p wt.
Effect of 80/81 and 24 Substitutions on the Cdc25p-mediated Activation of Ras2p wt on Adenylyl Cyclase-To study the effect of the various mutants on the Ras2p wt-dependent adenylyl cyclase activity, cell membranes were prepared from the mutated S. cerevisiae strain AAT3B-⌬1, carrying deletions in RAS1, RAS2, and CDC25 genes (Mirisola et al., 1994). The viability of this strain is ensured by the adenylyl cyclase CRI4 mutation that bypasses the requirement for RAS and CDC25 due to a higher intrinsic activity. Because the CRI4-encoded adenylyl cyclase activity is enhanced by the Ras protein much more than the wild type adenylyl cyclase (De Vendittis et al., 1986b), the sensitivity of the assay is increased. It is known that the addition of Ras2p⅐GDP to yeast membranes from this strain strain as a source of adenylyl cyclase is ineffective in stimulating the cAMP production . The presence of GEF restores the adenylyl cyclase activity via the conversion of Ras2p⅐GDP to Ras2p⅐GTP. Fig. 4 illustrates the capacity of the various Ras2p mutants to interfere with the action of Ras2p wt in a reconstituted system for adenylyl cyclase activity. Ras2p S24N hinders in a concentration-dependent manner the reconstitution of adenylyl cyclase activity dependent on the regeneration of Ras2p wt⅐GTP by Cdc25p-C. No difference in the inhibitory effect was observed between the 42-kDa native form and the 29-kDa C-terminal truncated form of Ras2p S24N, further confirming that the C-terminal region of Ras2p, at least if not post-translationally processed, is irrelevant for the interaction with the catalytic domain of GEF. Increasing amounts of the triple mutant did not reveal any interference with the activation of adenylyl cyclase dependent on the conversion of Ras2p wt⅐GDP to Ras2p wt⅐GTP by Cdc25p-C. As in a way expected, no interfering effect could be detected with Ras2p R80D/N81D. The moderate increase in cAMP production was due to the partial regeneration of the Ras2p R80D/N81D⅐GTP complex, because the GTP-and Gpp(NH)p-bound forms of this mutant can activate membrane-bound adenylyl cyclase (Verrotti et al., 1992;Mirisola et al., 1994). The observation that a five to six times higher concentration of its GDP complex was needed to induce half the Cdc25p-dependent adenylyl cyclase activity obtained with Ras2p wt⅐GDP is in agreement with the marked decrease in the affinity of the double mutant for GEF (cf. Table I).
From Fig. 4, one can derive that neither Ras2p S24N nor the triple mutant was able to stimulate the adenylyl cyclase activity. The lack of activity cannot be explained by a reduced affinity for GTP. Because the K d values of their GTP complexes lie in the 10 -90 M range (Table II), more than 50% of the bound GDP should be exchanged with GTP within a few seconds due to the high intrinsic GDP dissociation rates of these two mutants and the high concentration of GTP present in the assay (0.5 mM). In agreement with this, no adenylyl cyclase activity was detected with the preformed GTP complexes of these two Ras2p mutants.
To analyze whether the double and triple Ras2p mutants were still capable of interacting with adenylyl cyclase in an unproductive manner, we have carried out competition experiments between these mutants and Ras2p wt with respect to the activation of adenylyl cyclase. Neither Ras2p S24N nor Ras2p S24N/R80D/N81D in their GTP bound-form using a concentration excess up to 10 times over Ras2p wt⅐GTP could compete for the adenylyl cyclase activation. These results (not illustrated) show that the inability of these two Ras2p mutants to activate adenylyl cyclase is related to a defective binding to their target. DISCUSSION This work shows that the property of substitutions R80D/ N81D to abolish the interference on the Ras2p wt-GEF interaction induced by mutation S24N is associated with the loss of the ability to form a stable complex with GEF. The relationships between these two effects emphasize the importance of the distal switch II region for the binding to GEF. R80D/N81D and S24N are substitutions that induce opposite effects on the properties of Ras proteins. The former mutation has been reported to eliminate the response to GEF, affecting neither the intrinsic GDP/GTP exchange reaction of Ras2p nor its affinity for GDP and GTP (Verrotti et al., 1992;Poullet et al., 1995). In vivo, it has a lethal effect that can be rescued by a constitutively activating mutation that bypasses the need for GEF (Verrotti et al., 1992). Substitution S24N induces dominant negative properties similar to those described for the homologous p21 S17N (John et al., 1993) that are based on the formation of a stable complex with GEF, leading to the sequestration of the exchange factor. In this work, the characterization of Ras2p R80D/N81D has been extended by demonstrating that its insensitivity to GEF is caused by a strong decrease in the affinity between these two proteins. As an important feature, the introduction of R80D/N81D into Ras2p S24N does not influence two selective properties induced by the latter mutation: the strongly decreased affinity for GTP and GDP and the inability to activate adenylyl cyclase. Therefore, the suppression of the tight interaction with GEF virtually constitutes the only major difference between the triple mutant and Ras2p S24N.
The switch II region is known to be the structure of Ras undergoing the most extensive changes depending on whether GDP or GTP is bound, as shown by x-ray diffraction and NMR studies (Pai et al., 1990;Miller et al., 1992). The pivotal role of glycine 82 (Gly 75 in Ras p21), located at the boundary of the C-terminal end of helix ␣2 and loop L5, is crucial for the transitions between the GDP-and GTP-bound forms of Ras proteins (Stouten et al., 1993;Kavounis et al., 1991), strongly suggesting that the physiological action of GEF requires a specific conformation of loop L4/helix ␣2/loop L5 for inducing a productive interaction. The competition phenomena described between nucleotide and GEF for binding to Ras2p (Poullet et al., 1995; this work) could arise from the coordination existing between the helix ␣2 region and the nucleotide binding network (Stouten et al., 1993). In the dominant negative mutants of Ras, the binding to GEF, which in the case of the wild type Ras protein represents a transient intermediary state, becomes a stabilized specific state of the Ras molecule (Feig & Cooper, 1988;Farnworth & Feig, 1991). Interfering properties are induced by mutations of specific residues of the nucleotide binding site, particularly of those involved in a direct perturbation of the magnesium ion-nucleotide coordination (Farnsworth & Feig, 1991;Jung et al., 1994). As a common feature, these Ras mutants display a decrease in the affinity for GDP and GTP by several orders of magnitude. Because the nucleotide binding site is strictly correlated with the structural elements playing a key role in the interaction with ligands (effector loop) or in the  specific conformations determining the active and inactive states (switch I and II regions), it is not surprising that a single point substitution in this area can modify in a drastic manner the interaction with regulators and effectors. The relevant role of helix ␣2 in the binding of GEF appears to be associated not only with its boundary region to loop L5. Other surface-exposed residues of helix ␣2 in p21 (62, 63, 67, 69 (Mosteller et al., 1994), 66 (Howe & Marshall, 1993), and 69 (Segal et al., 1995), corresponding to Ras2p residues 69, 70, 74, 76, 73, and 76, respectively) were found in vivo or in vitro to influence negatively the interaction with GEF, likely by a decrease in the affinity between these two proteins. These observations are in agreement with the determinant role of the helix ␣2 region in the interaction with GEF. The finding that substitutions situated on the boundary between helix ␣3 and loop L7 in p21 also hinders the GEF signal (Segal et al., 1995) does not contrast with this conclusion. In fact, in the three-dimensional model of p21, the helix ␣2 region and the helix ␣3/loop L7 boundary are located on the nearly same exposed surface of the Ras molecule and are in contact (Stouten et al., 1993). Consequently, conformational changes in helix ␣2/loop L5 also influences the state of the helix ␣3/loop L7 region. This situation, together with the negative effect of their mutation on the GEF action, suggests that both regions are involved in the interaction with GEF, either directly as part of the Ras2p binding site for GEF or indirectly as key elements for inducing the active conformation of this binding site that is very probably located nearby. Unlike this, experimental evidence indicates that the phosphoryl binding loop 4, flanking the N-terminal end of helix ␣2, only participates in the transmission of the GEF signal, because point substitutions in this loop impair the GEF-dependent dissociation rate of the nucleotide without substantially decreasing the affinity for GEF (Mistou et al., 1992).
Despite the relevance of the conclusions derived from these studies, the precise nature of the structural elements delimiting the GEF binding surface between Ras and GEF remains as yet unclear. In fact, helix ␣2/loop L5 and helix ␣3/loop L7, the two most probable regions of Ras proteins involved in a direct interaction with GEF, have only been characterized by mutagenic analysis. More direct functional methods, such as competition experiments using oligopeptides corresponding to the Ras2p or p21 regions including the various ␣-helices and flanking sequences, have failed so far to locate a specific structure or structures of Ras proteins directly involved in the interaction with GEF. 4 Consequently, the functional findings still need to be confirmed by structural studies directed to unveil the threedimensional relationships of the Ras⅐GEF complex.
Unlike Ras2p S24N, whose GTP complex is incapable of activating adenylyl cyclase as already reported for the corresponding p21 mutant (Farnsworth et al., 1991), Ras2p R80D/ N81D⅐Gpp(NH)p was found to display the same affinity for adenylyl cyclase as Ras2p wt, only the V max of the adenylyl cyclase activation was somewhat decreased (Verrotti et al., 1992;Mirisola et al., 1994). These results, which in part differ with the conclusions derived from experiments using p21 mutants and yeast adenylyl cyclase (Segal et al., 1995), clearly indicate that the Ras2p binding site for GEF is distinct from the site activating the adenylyl cyclase. The lack of effect by the additional presence of R80D/N81D on the specific conformation induced by S24N concerning the activation of adenylyl cyclase, further confirms that the GEF signal originates from regions different from the binding site for the nucleotide and the coordinated magnesium ion. In this work we have demonstrated that the alteration of the properties of the triple and Ras2p S24N mutants with respect to adenylyl cyclase is based on a truly defective binding between these two proteins and not on the formation of an inactive complex.
In conclusion, the observations that substitution of serine 24, a residue of the nucleotide/magnesium ion binding site, can induce a stable interaction with GEF and that additional mutations in helix ␣2 abolish this interaction put emphasis on the close relationships of the helix ␣2/loop L5 region with the specific binding site for GEF. Ras2p S24N/R80D/N81D with GDP and GTP To determine the dissociation constants of the Ras2p⅐GTP complexes, 40 pmol of GDP-free Ras2p mutants, obtained as described (Créchet et al., 1990a), were incubated in 25 l of buffer A containing 10 mM MgCl 2 and 4 -80 M [␥-32 P]GTP (80 GBq⅐mmol Ϫ1 ]. After 4 min of incubation at 30°C, during which time no GTPase activity was measurable, 23-l aliquots were withdrawn and passed on nitrocellulose filter disks. To determine the dissociation constants of the Ras2p⅐GDP complexes, 5 pmol of GDP-free Ras2p mutants were incubated in 50 l of buffer A containing 10 mM MgCl 2 and 0.2 to 6 M [ 3 H]GDP (180 GBq⅐mmol Ϫ1 ). After 5 min of incubation at 30°C, 40-l aliquots were withdrawn and passed on nitrocellulose filter disks. The kЈ ϩ1 values were calculated from the ratio K d /kЈ ϩ1 using the values reported in Table I. NM, nonmeasurable with the nitrocellulose binding assay due to a dissociation rate that is too fast.  4. Ras2p S24N but not Ras2p R80D/N81D or Ras2p S24N/ R80D/N81D interferes with the adenylyl cyclase reconstituted activity dependent on Ras2p wt⅐GDP and Cdc25p-C. The 29 kDa (Ç) or 42 kDa (å, Ⅺ) form of Ras2p S24N, Ras2p R80D/N81D (E, q) or Ras2p S24N/R80D/N81D (f, ࡗ) in the indicated concentrations were incubated for 30 min at 30°C with 5 nM Cdc25p-C and 30 g of membrane from S. cerevisiae strain AAT3B-⌬1 in the presence (Ç, å, f, q) or the absence (E, ࡗ, Ⅺ) of 150 nM Ras2p wt⅐GDP. The reaction was started with a mixture containing the nucleotides GTP, cAMP, [␣-32 P]ATP (5 GBq⅐mmol Ϫ1 ), teophilline, creatine phosphate, creatine kinase, and MgCl 2 in 50 mM MES, pH 6.2 (see Materials and Methods). No adenylyl cyclase activity could be detected with the combination Cdc25p-C plus Ras2p S24N (Ⅺ) and Cdc25p-C plus Ras2p S24N/R80D/ N81D (ࡗ).