A Conserved Region of c-Ha-Ras Is Required for Efficient GTPase Stimulation by GTPase Activating Protein but Not Neurofibromin*

The effector binding domain and the switch II region of c-Ha-Ras are necessary for p120 GAP -stimulated GTP hydrolysis. We report a third region of c-Ha-Ras located within the (cid:97) 3 helix (amino acids 101–103) which is also required for efficient p120 GAP , but not neurofibromin-mediated hydrolysis. This highly conserved region of the Ras protein was investigated using an insertion-deletion mutant (Ras-100LIR104) originally character- ized by Willumsen et al. (Willumsen, B. M., Adari, H., Zhang, K., Papageorge, A. G., Stone, J. C., McCormick, F., and Lowy, D. R (1989) in The Guanine Nucleotide Binding Proteins; Common Structural and Functional Prop- erties (Bosch, L., Kraal, B., and Parmeggiani, A., eds) pp. 165–178, Plenum Press, New York). The 100LIR104 substitution did not alter the intrinsic hydrolytic rate of the protein. The p120 GAP -stimulated hydrolysis of Ras-100LIR104, however, was decreased by 2–3-fold com- pared to wild type Ras. This decrease in p120 GAP -stimu-lated hydrolysis was not due to its inability to physically associate with Ras-100LIR104 (cid:122) GTP (as determined by competitive binding assays). Surprisingly, neurofibro- min-stimulated GTP hydrolysis was unaltered by the mutation. Finally, no differences were observed in the ability of either the p120 GAP catalytic domain or the neurofibromin GRD to accelerate Ras-100LIR104

Ras proteins, which function in signal transduction pathways critical for cell growth and differentiation, are guanine nucleotide binding proteins with a M r of approximately 21,000. The membrane-associated Ras protein cycles between an inactive GDP-bound and active GTP-bound state. The activity of Ras is negatively regulated by the hydrolysis of bound GTP, and positively regulated by specific guanine nucleotide exchange factors which facilitate the replacement of bound GDP by GTP (Lowy and Willumsen, 1993). Growth factor stimulation of quiescent cells increases the proportion of Ras⅐GTP, resulting in the initiation of a cascade of intracellular protein kinases (Satoh et al., 1990;Zhang et al., 1992).
GTPase activating proteins negatively regulate Ras activity by accelerating the hydrolysis of bound GTP. Two such proteins have been extensively characterized, p120 GAP and neurofibromin (Lowy and Willumsen, 1993). The region of p120 GAP which is responsible for stimulating GTP hydrolysis is located in the carboxyl-terminal third of the protein and is termed the p120 GAP catalytic domain. The second GTPase activating protein for Ras, neurofibromin, is the product of the neurofibromatosis type-1 gene and has a M r of 250,000. Neurofibromin contains a 350-amino acid stretch located in the central portion of the protein which contains extensive sequence homology to the GAP 1 catalytic domain and is therefore termed the GAP related domain (GRD; Xu et al., 1990;Martin et al., 1990). Both the GAP catalytic domain and the neurofibromin GRD are highly active in accelerating the GTP hydrolysis by Ras.
Mutations in multiple regions of the Ras protein effect its interactions with p120 GAP and neurofibromin. Mutations in the Ras effector binding domain (amino acids 32-40) result in the inability of p120 GAP to enhance the intrinsic GTPase activity (Adari et al., 1988;Cales et al., 1988). Although not all mutations in this region abolish GAP activity, the effector region is thought to be essential for p120 GAP binding. In addition, analysis of Ras/Rap1A chimeras shows that deletion of residues 61-65 also renders Ras insensitive to p120 GAP stimulation (Zhang et al., 1991;Maruta et al., 1991). Many Ras mutations that are resistant to GAP stimulation have also been found to be resistant to stimulation by neurofibromin-GRD. Thus the effector binding domain (amino acids 32-40) and switch II region (amino acids 61-65) of Ras are thought to be the sites required for interaction with GTPase activating proteins. Recently a third GTPase activating protein has been identified. Its full characterization has not been reported (Maekawa et al., 1994).
In this paper we identify an additional region within the Ras protein which is important for p120 GAP -stimulated hydrolysis. This region of the Ras protein, consisting of amino acids 101-103, is highly conserved among a number of small GTP binding proteins (Santos and Nebreda, 1989). The functional significance of this highly conserved region of the Ras protein was investigated using an insertion-deletion mutant originally characterized by Willumsen et al. (1989). This mutation, in which the amino acids KRV were substituted for LIR at position 101-103 was designated Ras-100LIR104. The intrinsic hydrolytic activity of Ras-100LIR104 was unaltered compared to wild type Ras. The p120 GAP -stimulated hydrolysis by Ras-100LIR104, however, was reduced 2-3-fold, while neurofibromin hydrolytic activity remained unaffected. No differences were observed in the ability of either the p120 GAP catalytic domain or the neurofibromin GRD to accelerate Ras-100LIR104 GTPase activity, indicating that regions outside of the catalytic GAP domain are involved in protein-protein in-teractions with Ras necessary for efficient stimulation of GTP hydrolysis.
Purification of Ras and Ras-100LIR104 -p21 c-Ha-Ras and p21 c-Ha-Ras-100LIR104 (previously classified as mutant number 1742) were expressed in Escherichia coli. Luria broth (5 ml) containing 50 g/ml of ampicillian was inoculated from glycerol stocks of Ras or Ras-100LIR104 and incubated for 12 h at 37°C. The 5-ml culture was then added to 1 liter of Luria broth containing 50 g/ml of ampicillian and 100 M isopropyl-B-D-galactoside which was incubated overnight at 37°C. The 1-liter culture was chilled to 4°C and centrifuged at 5000 rpm for 30 min. The pellets were resuspended and centrifuged a second time at 5000 rpm for 30 min. The supernatant was removed and the pellets frozen on dry ice and stored at Ϫ80°C. The following day, pellets were resuspended in G-50 buffer (20 mM Tris-HCl, 1 mM MgCl 2 , 1 mM DTT, and 150 mM NaCl) containing 1 mg/ml of lysozyme, 1 mM DTT, 0.5 ng/ml of pepstatin A, 5 ng/ml of aprotinin, 5 ng/ml of leupeptin, and 1 mM phenylmethylsulfonyl fluoride followed by a 90-min incubation at 4°C. Subsequently, 2.5 g of DNase and 300 l of 4% deoxycholate were added, followed by a 75-min incubation at 4°C. The suspension was then centrifuged at 10,000 rpm for 30 min and the supernatant was concentrated in a Centriprep-10 filter and loaded on a 200-ml G-50 Sephadex column. Fractions were analyzed by SDS-polyacrylamide gel electrophoresis, stained with Coomassie Blue, and the Ras or Ras-100LIR104 containing fractions were pooled. The fractions of peak protein concentration were concentrated and again passed over the above gel filtration column. The Ras and Ras-100LIR104 protein were purified to approximately 95% homogeneity as revealed by examination of Coomassie Blue-stained SDS-polyacrylamide electrophoresis gels. Protein activity was determined using a GTP binding assay described below. Care was taken to ensure that the specific binding activity for GTP by each Ras protein preparation was equivalent before the preparation was used for the work described herein.
Purification of Neurofibromin-All purification steps were performed at 4°C. Rabbit brain cytosol was prepared in Buffer A (50 mM Tris, pH 7.5, 1 mM EDTA, 1 mM DTT) as described previously (Golubic et al., 1992). Twenty ml of rabbit brain cytosol at 18 -25 mg/ml containing 100 mM NaCl and 400 l of neurofibromin antibody were incubated for 3 h at 4°C. Swollen Protein A-Sepharose beads (600 l) were added to the lysate/antibody mixture and rotated on a mixing wheel at 4°C for 30 min. The Protein A-Sepharose beads were allowed to settle by gravity and were washed twice with buffer A (containing 0.2% Nonidet P-40 and 100 mM NaCl), twice with buffer A (containing 100 mM NaCl), once with buffer A, and finally with 100 mM Tris (pH 8.0). A 1-ml column was constructed and washed with 5 column volumes of 100 mM Tris-HCl (pH 8.0) and 10 column volumes of 10 mM Tris (pH 8.0). Neurofibromin was eluted with 100 mM glycine (pH 3.00). Approximately, 200-l aliquots were collected in tubes containing 50 l of 1 M Tris (pH 8.0) to neutralize the pH. Fractions were concentrated in a Centricon (1000 rpm) to a final volume of 500 l, followed by concentration in a Microcon (2000 rpm) to a final volume of 50 l. A final of 50% glycerol was added and 20-l aliquots of purified neurofibromin were store at Ϫ80°C. It should be noted that the purified neurofibromin protein remains bound to the antibody.
GTP Binding Assay-A 50-l reaction mixture containing 100 M GTP, 1 mM DTT, 10 mM EDTA, and 1 l of [␣-32 P]GTP (2 mCi/ml, 400 Ci/mmol) was incubated with different concentrations of Ras or Ras-100LIR104 protein. The reaction mixture was incubated at room temperature for 30 min and filtered through 0.45-m nitrocellulose filters. The filters were then washed with 5 ml of 25 mM Tris-HCl, 5 mM MgCl 2 , and 50 mM NaCl and counted using liquid scintillation spectrometry to determine the amount of [␣-32 P]GTP bound to Ras or Ras-100LIR104.
Immunoprecipitation of p120 GAP and Neurofibromin-In a 1.5-ml Eppendorf tube, 300 l of rabbit brain lysate (10 mg/ml) was incubated for 2 h at 4°C with antibodies specific for neurofibromin or p120 GAP . In a separate tube, 100 l of Protein A-Sepharose was washed 5 times with TBS ϩ Nonidet P-40 (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 0.1% Nonidet P-40). The prepared Protein A-Sepharose was then added to the lysate/antibody mixture and incubated for 30 min at 4°C. Samples were then spun in a microcentrifuge for 5 s to pellet the Protein A-Sepharose and remove the lysate. The Protein A-Sepharose was then washed 4 times with TBS ϩ Nonidet P-40. Two hundred l of TBS ϩ Nonidet P-40 were then added to the Protein A-Sepharose and the slurry was placed on a 1-ml sucrose cushion (1 M sucrose) and the sample was spun for 2 min in a microcentrifuge. The Protein A-Sepharose was then washed 3 times with 1 ϫ TBS ϩ Nonidet P-40 and 3 times with 20 mM Tris. The 100 l of Protein A-Sepharose was then diluted by 10-fold, mixed thoroughly, and designated volumes of the slurry were aliquoted (using a large mouthed pipette tip) for use in the GTPase filter binding assay described below. The 10-fold dilution insured that equal amounts of beads were reproducibly obtained.
GTPase Filter Binding Assay-A 70-l reaction mixture composed of 840 ng of active Ras or Ras-100LIR104, as measured by the GTP binding assay, was incubated with 1 mM EDTA, 1 mM DTT, 5 g of bovine serum albumin, 20 mM Tris-HCl (pH 7.5), and 3 l of [␥-32 P]GTP (30 Ci/mmol) for 10 min at 30°C. This 70-l reaction mixture was then added to a 700-l mixture containing 20 mM Tris-HCl (pH 7.5), 10 g of bovine serum albumin, 1 mM DTT, 1 mM GTP, and 10 mM MgCl 2 . A 50-l volume of the combined mixture was then added to tubes containing neurofibromin or p120 GAP activity for a total volume of 100 l and a final Ras concentration of 60 ng/tube (28 nM). The reaction mixtures were incubated for various lengths of time depending on the assay conditions. Following incubation, 80-l aliquots were filtered and washed with 5 ml of 25 mM Tris-HCl, 5 mM MgCl 2 , and 50 mM NaCl. The filters were then counted using liquid scintillation spectrometry. The amount of p120 GAP lysate or neurofibromin was chosen so that the assay was in the linear range. Thus, we based our measurements on protein activity rather than on protein concentration (which would include inactive p120 GAP or neurofibromin molecules).
Statistics-These analyses were extensively repeated and summarized statistically. Statistical analyses were, therefore, performed with the ANOVA model (by Lisa Rybicki, Department of Biostatistics and Epidemiology, Cleveland Clinic Foundation) which is similar to the t test, but allows comparisons of all the data obtained. This approach, therefore, uses all available data to obtain an estimate of the variability within the data. In the text and figure legends "p" values indicate the likelihood that the ratio of GTPase activation for Ras compared to Ras-100LIR104 is statistically different from 1.0 when all repetitions of a given experiment are considered (p Ͻ 0.05 is significant). The ratios reported represent the ratio of GTPase stimulation with Ras-100LIR104 compared to GTPase stimulation with wild type Ras, reported as a range of ratios having a confidence level of 95%. "n" represents the total number of determinations in at least two separate experiments.
Competitive Binding Assay; Reagent Preparation-Ras was first allowed to associate with [␥-32 P]GTP: 128 pmol of Ras was incubated with 10 pmol of [␥-32 P]GTP (30 Ci/mmol), 2 mM EDTA, 2 mM DTT, 50 mM Tris-HCl (pH 7.5), and 100 g of bovine serum albumin in a 500-l reaction volume for 15 min at 30°C. Immediately following the incubation, 5 mM MgCl 2 was added to the reaction mixture which was then separated on a 9-ml G25 column equilibrated with 20 mM HEPES (pH 7.5) and 5 mM MgCl 2 . The column was run at 4°C with 1-ml fractions collected. Column fractions of 20 l were counted directly using liquid scintillation spectrometry and another 20 l was analyzed by a filter binding assay in order to distinguish Ras bound [␥-32 P]GTP from Ras fractions containing free [␥-32 P]GTP. Fraction 4 yielded 80% bound Ras⅐GTP (6 nM) which was aliquoted and frozen at Ϫ70°C. For binding Ras or Ras-100LIR104 with GMP-PNP between 3 and 6 mg of Ras or Ras-100LIR104, 2.5 mM GMP-PNP, 1 mM DTT, and 5 mM EDTA were incubated at 30°C for 20 min. Following incubation, 10 mM MgCl 2 was added and the reaction volumes were diluted (approximately 5-fold) with 20 mM HEPES (pH 7.5) and 1 mM MgCl 2 . Reaction volumes were concentrated in a Centriprep-10 (3500 rpm) after the addition of 20 mM HEPES (pH 7.5) and 1 mM MgCl 2 . Aliquots were frozen at Ϫ70°C. The final concentrations of Ras⅐GMP-PNP and Ras-100LIR104⅐GMP-PNP were between 2 and 6 mM as determined by a GTP binding assay.
Competitive Binding Assay-Baculovirus p120 GAP lysate or affinity purified neurofibromin (the amount used yielded 50% Ras⅐GTP hydrolysis) and increasing concentrations of Ras⅐GMP-PNP or Ras-100LIR104⅐GMP-PNP in 20 mM HEPES, 1 mM MgCl 2 , and 1% bovine serum albumin were prewarmed at 27°C for 2 min and incubated with 6 nM Ras⅐[␥-32 P]GTP for 4 min. The amount of p120 GAP lysate or neurofibromin sufficient to yield 50% hydrolysis was chosen so that the assay was in the linear range. Thus, we based our measurements on protein activity rather than on protein concentration (which would include inactive p120 GAP or neurofibromin molecules). The reaction was quenched with 0.5 ml of 4% charcoal in 50 mM NaH 2 PO 4 and spun in a Microfuge for 3 min. Two hundred l of supernatants were counted for 10 min to measure the release of 32 P i . Note that the 4-min time points were achieved by staggering the addition of Ras⅐[␥-32 P]GTP with the addition of 0.5 ml of 4% charcoal. Intrinsic hydrolysis during this time was minimal.

RESULTS
The Intrinsic Hydrolytic Activity of Ras and Ras-100LIR104 -To investigate whether the 100LIR104 substitution would alter the intrinsic hydrolytic activity of the protein, wild type Ras and Ras-100LIR104 were compared using a GTPase filter binding assay. It is important to note that when equal concentrations of the two proteins were used, wild type Ras and the Ras-100LIR104 mutant were able to bind equal amounts of [␣-32 P]GTP (data not shown). Therefore, both Ras and Ras-100LIR104 expressed the same GTP binding activity. To measure intrinsic hydrolytic activity, identical concentrations (28 nM) of wild type Ras or Ras-100LIR104 were individually allowed to associate with [␥-32 P]GTP, and then incubated at 30°C for the various times. Hydrolysis of bound [␥-32 P]GTP would result in the loss of protein-associated phosphate label. Following incubation, the amount of labeled phosphate remaining associated with Ras protein was quantitated by passing the samples over nitrocellulose filters (which retained the Ras protein and associated labeled phosphate) followed by liquid scintillation spectrometry. Both wild type Ras⅐GTP and Ras-100LIR104⅐GTP possessed the same intrinsic rate of GTP hydrolysis (Fig. 1).
Stimulation of Ras and Ras-100LIR104 GTPase Activity-In order to separately characterize the interaction of the mutant protein with the two GTPase activating proteins, neurofibromin and p120 GAP had to be physically separated from each other. This was initially accomplished by immunoprecipitation. Antibodies specific for p120 GAP or neurofibromin were incubated with 300 l of rabbit brain lysate (which contains both GTPase activating proteins) followed by the addition of Protein A-Sepharose to bind the immune complexes. The activity of the immunoprecipitated proteins was assayed by incubating increasing amounts of either p120 GAP or neurofibromin bound to Protein A-Sepharose with Ras⅐[␥-32 P]GTP or Ras-100LIR104⅐[␥-32 P]GTP for 45 min at 30°C. The stimulatory effects of immunoprecipitated p120 GAP or neurofibromin on the intrinsic hydrolytic rates of these two proteins were assessed using the filter binding assay previously described. No differences in GTPase stimulation were observed when increasing amounts of neurofibromin-bound Protein A-Sepharose was incubated with Ras⅐[␥-32 P]GTP compared to Ras-100LIR104⅐[␥-32 P]GTP (Fig. 2a). The ability of neurofibromin to stimulate GTPase activity was therefore not altered by the Ras mutation. However, when p120 GAP -bound Protein A-Sepharose was incubated with Ras⅐[␥-32 P]GTP, the stimulation of GTP hydrolysis was greater than when p120 GAP was incubated with Ras-100LIR104⅐[␥-32 P]GTP at all concentrations tested (Fig. 2b). The GTPase activity of p120 GAP with the mutant was reduced approximately 2-3-fold compared to wild type Ras.
These results were repeated multiple times and the data collectively analyzed using the methods described under "Experimental Procedures." The probability that neurofibromin stimulated the GTPase activity of wild type Ras and mutant Ras to different extents was not significant (p ϭ 0.5); while the probability that p120 GAP stimulated wild type Ras and mutant Ras differently was highly significant (p ϭ 0.0012).
In order to confirm the above results, neurofibromin and p120 GAP proteins were separated from each other utilizing an entirely different approach. Neurofibromin was affinity purified by passage of rabbit brain lysate over a Protein A-Sepharose column bound to neurofibromin-specific antibodies. The neurofibromin-antibody complex was then released from the Protein A-Sepharose. The neurofibromin thus purified ap-FIG. 1. Intrinsic rate of GTP hydrolysis by Ras and Ras-100LIR104. Ras or Ras-100LIR104 were allowed to associate with [␥-32 P]GTP and then incubated at 30°C for the indicated lengths of time. The extent of hydrolysis was calculated after incubation by determining the amount of label remaining associated with Ras protein in a filter binding assay. The % hydrolysis was determined as the ratio of binding at zero time to the reduction in the binding after incubation. In this and each of the experiments described, care was taken to ensure that the extent of GTP hydrolysis within a given experiment was between approximately 20 and 80% hydrolysis so that the extent of hydrolysis could be roughly equated to the amount of GTPase activation.

FIG. 2. Hydrolysis by immunoprecipitated GAP and neurofibromin (NF).
Neurofibromin and p120 GAP were separately immunoprecipitated from rabbit brain lysate and bound to Protein A-Sepharose. The indicated volumes of this immobilized immunoprecipitate were incubated for 40 min at 30°C with Ras or Ras-100LIR104 bound to [␥-32 P]GTP. The GTPase filter binding assay was then used to determine hydrolysis (see Fig. 1). Data is representative of three experiments: (a) immunoprecipitated neurofibromin; and (b) immunoprecipitated p120 GAP . The data presented were obtained in a single experiment, while three separate experiments were analyzed together statistically as described under "Experimental Procedures" yielding the following results: for p120 GAP , n ϭ 9, p ϭ 0.0012, ratio is 1.8 -2.6; and for neurofibromin, n ϭ 12, p ϭ 0.5, ratio is 0.9 -1.1.
peared as a single band on a silver-stained gel (together with the immunoglobulin, data not shown). The p120 GAP was prepared from a recombinant baculovirus expression system (from R. Jove: Park et al., 1992). Lysates of baculovirus expressing cells were used as a source of p120 GAP . These lysates had no endogenous GTPase activating activity in the presence of Ras unless the p120 GAP protein had been expressed.
With a GTPase filter binding assay, the affinity-purified neurofibromin protein was found to stimulate Ras-100LIR104 hydrolysis to the same extent as with wild type Ras (Fig. 3a). In contrast, the ability of baculovirus p120 GAP to stimulate the intrinsic hydrolysis of Ras-100LIR104 was reduced by 2-3-fold compared to wild type protein (Fig. 3b). These conclusions are highly significant when all results are statistically analyzed (see figure legend for statistical data). These results were similar to the differences in activity observed using p120 GAP and neurofibromin immunoprecipitated from rabbit brain lysate. It is also important to note that the GTPase filter binding assays discussed above were also performed at Ras or 100LIR104 Ras concentrations of 12 M, the concentration of Ras required to bind 50% of p120 GAP molecules. At this Ras concentration the same decrease in baculovirus p120 GAP -stimulated hydrolysis of 100LIR104 Ras compared to wild type Ras was observed (data not shown), indicating that the differences seen are not a result of low Ras concentrations. In summary, the results obtained using affinity purified neurofibromin and baculovirus p120 GAP accurately reflect those observed in the previous determina-tion, and indicate that only p120 GAP has reduced activity with the mutant Ras protein.
A final experiment was performed to confirm that p120 GAP (and not neurofibromin) has reduced ability to stimulate the GTPase activity of Ras-100LIR104. The detergent n-dodecylmaltoside has been shown to specifically inactivate neurofibromin in brain lysates (Bollag and McCormick, 1991). When a rat brain lysate was treated with maltoside to neutralize the neurofibromin activity, a 2-3-fold decrease in its ability to stimulate GTPase was evident with the Ras-100LIR104 mutant compared to wild type Ras, similar to results obtained above. In untreated brain lysates on the other hand, where a preponderance of the GTPase activating activity is due to neurofibromin, the decrease in GTPase stimulation of Ras-100LIR104 was not observed (data not shown; see "Discussion" for details).
Because of the difficulty in obtaining purified neurofibromin or p120 GAP , together with the loss of activity of these two reagents experienced during purification, we elected to rapidly separate the two activities from each other using three entirely separate approaches. With each of these approaches almost identical results were obtained indicating that the 101-103 region is required for efficient p120 GAP stimulated hydrolytic activity of Ras; but not for neurofibromin-stimulated hydrolysis. While the means of separating p120 GAP from neurofibromin differed in each experimental approach, the similarity in results obtained with each, together with the high degree of reproducibility of the data, strongly argue in favor of the above conclusion.
Activities of the p120 GAP Catalytic Domain and the Neurofibromin GRD-Attempts were next made to determine which part of the p120 GAP molecule was responsible for the decreased interaction seen with the mutant Ras protein. While p120 GAP and neurofibromin are largely unrelated to each other, the catalytic domain of p120 GAP and the GRD of neurofibromin display similar activities and share approximately 26% sequence homology (Martin et al., 1990). These peptide regions, each of which possess the ability to efficiently induce GTPase activation of Ras proteins, have been separately expressed (Hettich and Marshall, 1994;Xu et al., 1990). The p120 GAP catalytic domain and the neurofibromin GRD were purified as glutathione S-transferase fusion proteins produced from bacterial vectors and purified by affinity chromatography. These two proteins were then incubated with Ras or Ras-100LIR104 in the previously described GTPase assay. No difference in the neurofibromin GRD activity was observed when Ras and Ras-100LIR104 were compared (Fig. 4a), similar to observations with full-length neurofibromin. Furthermore, the catalytic carboxyl-terminal portion of p120 GAP was equally effective in its ability to accelerate wild type and Ras-100LIR104 GTPase activity (Fig. 4b). As above, these conclusions are based upon statistical analyses of all the data obtained from several separate experiments (see figure legend for statistical summary). The mutant Ras protein therefore retains the ability to be stimulated enzymatically by the catalytic core of p120 GAP . The reduced stimulation observed with full-length p120 GAP must therefore result from an altered interaction between the amino terminus of p120 GAP and the altered structure of the mutant Ras. Obviously this interaction is important in controlling p120 GAP activity.
Determination of Binding Affinities-The question then remains, does the inability of full-length p120 GAP to efficiently stimulate the GTPase activity of Ras-100LIR104 result from the inability of p120 GAP to bind to the mutant protein, or to its inability to enzymatically accelerate its GTPase activity? Competitive binding assays were preformed to distinguish between these two possibilities. This assay is designed to determine how FIG. 3. Hydrolysis by separately purified GTPase activating proteins. a, neurofibromin (NF1) was affinity purified from rabbit brain lysate and the indicated volumes of purified protein were incubated with Ras⅐GTP or Ras-100LIR104⅐GTP for 30 min at 30°C. b, the indicated volumes of baculovirus p120 GAP cell lysate were incubated with Ras⅐GTP or Ras-100LIR104⅐GTP for 30 min at 30°C. Reaction mixtures were then analyzed using a GTPase filter binding assay. (Statistical analysis of data from two separate experiments yielded for baculovirus GAP, n ϭ 14, p ϭ 0.0001, ratio 3.2-4.6; and for purified neurofibromin, n ϭ 9, p ϭ 0.1, ratio 0.8 -1.0.) much Ras protein bound to an unlabeled GTP analogue is required to competitively inhibit GTPase stimulation of a trace amount of Ras bound to labeled GTP. These assays were, therefore, performed with a low concentration (6 nM) of labeled, wild type Ras⅐[␥-32 P]GTP incubated with baculovirus p120 GAP or affinity purified neurofibromin. The extent of hydrolysis was then determined with a charcoal binding assay in the presence of increasing concentrations of competitive Ras protein bound to an unlabeled, nonhydrolyzable GTP analogue (GMP-PMP). When concentrations of the unlabeled, competitive protein reach the binding constant, hydrolysis of the labeled GTP bound to wild type Ras would be reduced 50% (IC 50 ). The binding affinity of neurofibromin for Ras and Ras-100LIR104 were identical (IC 50 ϭ 0.030 M, Fig. 5a). Similarly, the binding of p120 GAP to Ras and Ras-100LIR104 was also equal (IC 50 ϭ 12 M, Fig. 5b). The values obtained here for both neurofibromin and p120 GAP are similar to those previously reported (Bollag and McCormick, 1991).
These experiments rule out the possibility that the decrease in p120 GAP activity in the presence of Ras-100LIR104 results from a decreased binding affinity. The decrease in GTPase stimulation of Ras-100LIR104, therefore, is due to the inability of full-length p120 GAP to efficiently stimulate the hydrolysis of the mutant protein once it has bound. Clearly, amino acids in the region of 101-103 of Ras are important for the interaction between p120 GAP and Ras proteins, an interaction which apparently is important for full activity of p120 GAP . It is interesting, however, that this region is apparently not involved in the interaction between Ras and neurofibromin. It is possible that the amino-terminal regions of p120 GAP (which are necessary to achieve efficient GTPase activity, and which are totally distinct from any sequence in neurofibromin) physically interact with the carboxyl terminus of the ␣3 region of Ras or another region of Ras altered by the 101-103 mutation.
Lastly, In order to address the biological significance of the Ras-100LIR104 mutation, wild type Ras or the Ras-100LIR104 mutant were microinjected into NIH3T3 cells. The cells were subsequently labeled with [ 3 H]thymidine. The results indicated that there was no difference in the ability of the cell microinjected with Ras or Ras-100LIR104 to enter S phase (data not shown). In the future, experiments designed to transfect different cell types with Ras-100LIR104 may provide further insight regarding the biological role of this mutant.

DISCUSSION
The amino acids KRV located in the ␣3 helix of Ras at position 101-103 are highly conserved in Ras and several other small GTP binding proteins (Santos and Nebreda, 1989). This conserved region of the Ras protein was investigated using an insertion-deletion mutant Ras-100LIR104. The 100LIR104 substitution does not alter the intrinsic GTP binding or hydrol- FIG. 4. GTPase activation by catalytic domains. a, the indicated volumes of purified neurofibromin GRD were incubated with Ras⅐GTP or Ras-100LIR104⅐GTP for 30 min at 30°C. b, the indicated volumes of purified p120 GAP catalytic domain were incubated with Ras⅐GTP or Ras-100LIR104⅐GTP for 30 min at 30°C. Reaction mixtures were then analyzed using a GTPase filter binding assay. (Statistical analysis of data from three separate experiments yielded for GAP catalytic domain, n ϭ 13, p ϭ 0.50, ratio 1.0 -1.1; while for the GRD of neurofibromin, n ϭ 15, p ϭ 0.5, ratio 0.8 -1.0.) FIG. 5. Determination of binding affinities. Binding affinities were determined by a competitive inhibition approach. Ras protein was allowed to bind [␥-32 P]GTP and the unbound label removed by gel filtration. Then, 6 nM of the labeled Ras was incubated 4 min with enough (a) affinity purified neurofibromin or (b) baculovirus p120 GAP to induce approximately 50% hydrolysis. In these determinations the extent of hydrolysis was determined by a charcoal binding assay (see "Experimental Procedures"). This hydrolysis was then assayed in the presence of the indicated increasing concentrations of either Ras or Ras-100LIR104 which had been allowed to bind a nonhydrolyzable analogue of GTP. The concentration of unlabeled competitor which results in a 50% decrease in the extent of hydrolysis is the IC 50 . Data is representative of three experiments. ysis of the protein but does diminish p120 GAP -stimulated hydrolysis by 2-3-fold. Neurofibromin hydrolytic activity, however, was not altered in Ras-100LIR104 when compared to wild type Ras. This result was obtained regardless of the means by which p120 GAP and neurofibromin were separated from each other. A biochemical purification of these two GTPase activating proteins was not attempted due to the extensive manipulations required and the consequent loss and potential alteration of their activities. Three methods were therefore utilized to rapidly separate the two proteins and assess their abilities to stimulate the GTPase activity of mutant and wild type proteins. First, the two proteins were separately immunoprecipitated and assayed directly. Second, the neurofibromin was affinity purified on an antibody column and the neurofibrominimmunoglobulin complex released and utilized in comparison to a lysate of baculovirus expressed p120 GAP . Finally, crude brain lysate was assayed after the neurofibromin activity had been specifically inhibited by a detergent.
In the first instance the two proteins produced in mammalian cells were isolated as immunoprecipitates. In the second approach the p120 GAP was produced in an insect cell where no endogenous GTPase activity has been observed (data not shown). In this case the protein would be free of other mammalian proteins. In the third procedure the p120 GAP was assayed in the presence of other soluble mammalian proteins, but with inactivated neurofibromin. The fact that essentially identical results were obtained in repeated analyses of each type clearly indicate that there is a distinction in the interaction of these two GTPase activating proteins with the mutant Ras. On the other hand, the catalytic domain of p120 GAP and the GRD of neurofibromin were found to stimulate the rate of GTP hydrolysis of Ras and Ras-100LIR104 efficiently, and to the same extent. It is therefore concluded that an interaction between the amino-terminal, noncatalytic region of p120 GAP and the 101-103 position of Ras must be essential for efficient GTPase stimulation, while no such interactions are required for the activity of neurofibromin. Alternatively, alterations in the structure of distant regions of Ras resulting from the mutation might also play a role.
In an attempt to address the biological significance of the Ras-100LIR104 mutation, this protein was microinjected into NIH3T3 cells. We found, however, no difference in the ability of the cells microinjected with Ras or Ras-100LIR104 to enter S phase (data not shown). These results are, however, not unexpected when considering our data obtained using rat brain lysate. When rat brain lysate was used as a source of GAP (p120 GAP and neurofibromin) activity, no difference was observed between Ras and the Ras-100LIR104 mutant. It was only the separation of p120 GAP from neurofibromin which resulted in the identification of a decrease in p120 GAP activity in the presence of Ras-100LIR104. It is therefore conceivable that microinjection of the Ras-100LIR104 mutant into cells containing both p120 GAP and neurofibromin, would not produce different effects. Additional experiments in which the Ras-100LIR104 mutant is transfected into different cell types, however, may further our understanding regarding the biological significance of this mutation.
It has been previously reported that the Ras-100LIR104 mutation exhibits normal sensitivity to p120 GAP . However, the source of GAP in these studies was an MCF-7 cell extract which would be expected to contain both p120 GAP and neurofibromin (Adari et al., 1988). Those experiments were performed at a Ras concentration of 0.8 M. Since the binding constant of Ras for neurofibromin and p120 GAP is 30 nM and 12 M, respectively, a Ras concentration of 0.8 M would result predominantly in neurofibromin-mediated Ras⅐GTP hydrolysis. Thus, it is under-standable that no decrease in GTP hydrolysis was detected when Ras-100LIR104 was incubated with MCF-7 cell lysate. For the same reason we also did not observe differences in GTP hydrolysis between Ras and Ras-100LIR104 when rabbit brain lysate (which contains both neurofibromin and p120 GAP ) was used as a source of GAP (data not shown). Only when steps were taken to separate the p120 GAP and neurofibromin activity were the differences seen.
The ␣3 helix of Ras has been implicated in other studies as a site for GAP interaction (Wood et al., 1994). The yeast RAS2-E99K mutation which exhibited intrinsic hydrolytic rates similar to wild type Ras, showed reduced sensitivity to three separate GTPase proteins: IRA2 GAP, E. coli expressed mammalian GAP, and neurofibromin. Furthermore, Wood et al. (1994) also showed by a competitive binding assay that the affinity of RAS2-E99K for neurofibromin was dramatically reduced. These results differ from ours, where neither the neurofibromin activity nor binding affinity for Ras-100LIR104 was diminished. The yeast RAS2-E99K mutation would be located at the beginning of the ␣3 helix of the protein, while the 100LIR104 mutation is located at the terminus of this helix. While the individual characteristics of the two types of mutants differ, both result in clear alterations in interactions with GTPase activating proteins supporting the conclusion that this region of Ras, in addition to the effector binding domain and residues 61-65 of the switch II region, is an important site for interaction with GTPase activating proteins. It is also interesting to note that the effector domain, the switch II region, and a region of the ␣3 helix, all reside on the same surface of the Ras protein. These three regions are in close proximity to one another and could potentially provide a binding site for GAP proteins (Wood et al., 1994). Finally, the Ras-D38E mutant is not activated by p120 GAP but is able to bind to p120 GAP with an affinity similar to wild type. Interestingly, the D38E mutation causes a chain shift in the 101-103 region of Ras. Thus the D38E mutation may have low GTPase activity in part due to the chain shift in the 101-103 region of Ras (Krengel et al., 1990). Alternatively, both this mutation and the Ras-100LIR104 mutation might induce conformational changes at other sites of the protein which are important in modulating GTPase activation.
Other studies have also examined Ras-100LIR104 in the presence of GAP purified from placenta. Placental GAP, designated p100 GAP , is generated from an alternative splicing product and encodes a protein product with a predicted molecular weight of 100,400 (Trahey et al., 1988). p100 GAP lacks the hydrophobic amino terminus of the p120 GAP species, but retains catalytic activity. No difference in hydrolysis was observed when Ras and Ras-100LIR104 were assayed for p100 GAP activity (Downward et al., 1990). This observation is consistent with our data where decreased hydrolysis was observed with Ras-100LIR104 only in the presence of full-length p120 GAP . Therefore, the hydrophobic, extreme amino terminus of p120 GAP appears to be responsible for the differences in GTPase activation observed between Ras-100LIR104 and wild type Ras. While it is not clear exactly how Ras protein is positioned with respect to the plasma membrane, it is likely that the protein is oriented with the lipid-modified carboxyl terminus oriented toward the membrane. If so, it is further likely that the terminal region of the ␣3 helix of Ras containing amino acids 101-103 is located near the plasma membrane (de Vos et al., 1988;Pai et al., 1989). If, as postulated above, the amino terminus of p120 GAP interacts with the terminus of the ␣3 region of Ras, this hydrophobic region of p120 GAP would be expected to be positioned near, or perhaps even at, the plasma membrane. It is additionally clear from these studies that this region of p120 GAP plays a critical role in modulating GTPase activation.
Recently, a novel mammalian GTPase activating protein for Ras, Gap1 m , has been identified (Maekawa et al., 1994). Because Gap1 m is expressed in brain, it is possible (in those experiments in which rabbit brain lysate was used as a source of p120 GAP ) that Gap1 m function is also impaired in the presence of the Ras-100LIR104 mutant. Further investigation, however, is needed in order to determine the interaction of Gap1 m with 100LIR104 Ras.
It is also possible that associated protein(s) are in a complex with p120 GAP . This possibility, however, does not diminish the importance of the observation that the conserved 101-103 amino acids in the ␣3 helix of Ras are specifically involved in p120 GAP -mediated hydrolysis but not neurofibromin-mediated hydrolysis. The likelihood that associated proteins are bound to p120 GAP is lessened by the fact that p120 GAP is overexpressed in insect cells, therefore, the ratio of an associated protein to overexpressed p120 GAP would differ significantly to that observed in rat brain lysate in which p120 GAP is not overexpressed. Thus, if an associated protein were involved, one would expect different results from p120 GAP overexpressed in baculovirus lysates compared to endogenous p120 GAP in rat brain lysates. Since identical results were obtained in both cases, it is unlikely that an associated protein contributes significantly to the decrease in GTPase activity.
Another interesting possibility is that the decrease in p120 GAP activity is not due to the deletion of the conserved KRV amino acids but the substitution of amino acids LIR. Future experiments, in which single substitutions or a deletion of the one or all three of the conserved amino acids may help us better answer this question or perhaps even lead to the identification of additional mutants of this type.
Lastly, the Ras-100LIR104 mutant may serve as an important tool to better understand the function of the GAP proteins. For example, such a mutant may differentiate between different GAPs or between neurofibromin and p120 GAP . The construction of cell lines transfected with the Ras-100LIR104 mutation could thus provide a useful in vivo approach to address these types of questions.