Ras Isoforms Vary in Their Ability to Activate Raf-1 and Phosphoinositide 3-Kinase*

Ha-, N-, and Ki-Ras are ubiquitously expressed in mammalian cells and can all interact with the same set of effector proteins. We show here, however, that in vivo there are marked quantitative differences in the ability of Ki- and Ha-Ras to activate Raf-1 and phosphoinositide 3-kinase. Thus, Ki-Ras both recruits Raf-1 to the plasma membrane more efficiently than Ha-Ras and is a more potent activator of membrane-recruited Raf-1 than Ha-Ras. In contrast, Ha-Ras is a more potent activator of phosphoinositide 3-kinase than Ki-Ras. Interestingly, the ability of Ha-Ras to recruit Raf-1 to the plasma membrane is significantly increased when the Ha-Ras hypervariable region is shortened so that the spacing of the Ha-Ras GTPase domains from the inner surface of the plasma membrane mimicks that of Ki-Ras. Importantly, these data show for the first time that the activation of different Ras isoforms can have distinct biochemical consequences for the cell. The mutation of specific Ras isoforms in different human tumors can, therefore, also be rationalized.

Ras proteins operate as molecular switches in signal transduction pathways downstream of tyrosine kinases. An interesting yet unresolved issue is whether the Ras isoforms that are ubiquitously expressed in mammalian cells serve distinct functions. Several lines of evidence suggest that they may. First, recent studies have shown that Ki-Ras but not Ha-or N-Ras has an essential function in mouse development (1,2). Secondly, specific Ras isoforms are mutated in different tumors: Ki-ras mutations occur in 50% of colon cancers and 90% of pancreatic cancers, whereas N-and Ha-ras mutations are extremely uncommon. Conversely, N-ras mutations occur in 25% of acute leukemias, whereas Ha-ras and Ki-ras mutations are much less common (3). The simplest interpretation of these observations is that oncogenic activation of different Ras proteins has distinct biological consequences for the cell.
The N-terminal 165 amino acids of Ras contain all of the critical domains for GTPase function. The N-terminal 85 residues of all Ras isoforms are identical and contain the two switch regions that undergo conformational changes on GTP binding; in addition the next 80 amino acids are 95% conserved. All Ras proteins in the activated GTP bound state interact with the same set of effectors: Raf kinases, phosphoinositide 3-ki-nase (PI3-K), 1 RalGDS, and AF6 (4 -10). Mutational and structural studies have demonstrated that the Ras effector domain (residues 32-40) is a critical binding site for all these Ras effectors. Moreover, where measured using recombinant proteins, no marked differences in binding affinities of Ha-, N-, and Ki-Ras for these various effector proteins are apparent (11). Thus, there are no differences in effector domains or flanking sequences that could account for distinct signaling outcomes downstream of each Ras isoform.
The major differences between Ras proteins are confined to the hypervariable region (HVR) between residues 166 and 185.
Here the Ras sequences diverge significantly with fewer than 15% conserved residues (12). C-terminal to the HVR all Ras proteins terminate in a conserved CAAX motif (C, cysteine; A, aliphatic amino acid; X, methionine or serine) that directs post-translational processing. In previous studies we have shown that one role of the HVR is to cooperate with the processed C-terminal CAAX motif and provide a second signal for Ras plasma membrane localization (13,14). In Ki-Ras this second signal comprises a polylysine domain (lysine residues 175-180), whereas in Ha-Ras and N-Ras the second signal comprises palmitoylation sites at cysteines 181 and/or 184 (15).
Ras must be localized to the inner surface of the plasma membrane to be biologically active. In addition, Raf-1 and PI3-K are constitutively activated when targeted to the plasma membrane using Ras localization motifs (16 -18), indicating that the recruitment of these effectors to the plasma membrane by Ras is a critical step for activation. Taken together, these data suggested that the ability of Ras proteins to activate specific effectors may be influenced by their different mechanisms of attachment to the plasma membrane. To test this hypothesis we investigated whether Ki-Ras and Ha-Ras, which use markedly different plasma membrane localization signals, are equipotent activators of Raf-1 and PI-3K.

MATERIALS AND METHODS
Cell Fractionation and Immunoblotting-COS cells were electroporated as described previously (19) using 1-30 g of EXV-Ki-RasG12V, EXV-Ha-RasG12V, or EXV-Ha-Ras⌬G12V expression plasmids. 72 h later, after an 18-h serum-free incubation, cells were washed, scraped on ice into 0.5 ml of Buffer A (10 mM Tris-Cl, pH 7.5, 25 mM NaF, 5 mM MgCl 2 , 1 mM EGTA, 1 mM DTT, and 100 M NaVO 4 ), and homogenized in a tight fitting Dounce homogenizer. The postnuclear supernatants were spun at 100,000 ϫ g, the supernatants (S100) were removed, and the sedimented fractions (P100) were rinsed and sonicated for 5 min in 100 l of ice-cold Buffer A. Protein content was measured by the Bradford reaction. Expression of Ras was determined by quantitative immunoblotting. 20 g of each P100 fraction was resolved on 15% * This work was supported by a grant from the National Health and Medical Research Council (Australia). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by the Royal Children's Hospital Foundation (Queensland, Australia). To whom correspondence should be addressed: Dept. of Pathology, University of Queensland Medical School, Herston Rd., Brisbane 4006, Australia. Tel.: 61-7-3365-5340; Fax: 61-7-3365-5511; Email: j.hancock@mailbox.uq.edu.au. SDS-PAGE gels and transferred to polyvinylidene difluoridepolyvinylidene difluoride membranes using semidry transfer. Aliquots of GST-Ras, ranging from 5-400 ng (containing therefore 2.1-171 ng of Ras protein) were also loaded onto gels and transferred onto the same polyvinylidene difluoride membrane as the P100 fractions. Westerns were probed with anti-Ras (Y13-259) and then anti-Rat horseradish peroxidase (Pierce) antibodies, developed using enhanced chemiluminescence (SuperSignal, Pierce) and quantitated by phosphorimaging with a CH screen (Bio-Rad). The signals from GST-Ras were used to construct a dose response curve from which the actual Ras content of each P100 fraction was determined. For co-transfections 10 g of EXV-FLAGRaf plasmid (20) was electroporated with 1-30 g of Ras plasmid. The cells were harvested and Ras content in 20 g of P100 was determined by immunoblotting as described. 20 g of each S100 and P100 fraction were immunoblotted for FLAGRaf with M2 anti-FLAG monoclonal (Kodak) and anti-mouse horseradish peroxidase (Pierce). FLAG immunoblots were quantitated by phosphorimaging and the percentage of Raf-1 recruited to the P100 fraction calculated as P100⅐Raf/(P100⅐Raf ϩ S100⅐Raf).
Raf-1 Membrane Kinase Assays-This assay is discussed in detail elsewhere (20). Briefly, P100 aliquots (20 g) were adjusted to 20 l with Buffer A. 2.2 l of 10% Nonidet P-40 was added, and the membranes were sonicated in a sonicating waterbath for 2 min at 4°C. A 10-l aliquot of sonicated P100 fraction was incubated with 6 l of Buffer A containing 0.25 g of MEK, 1 g of ERK, and 4 l of 0.5 mM ATP/40 mM MgCl 2 and vortexed at 30°C. A second 10-l aliquot of sonicated P100 fraction was incubated with 6 l of Buffer A containing 4 l of 0.5 mM ATP/40 mM MgCl 2 , 1 g of ERK (but no MEK), and the mix was vortexed at 30°C (which is the control tube). After 20 min the samples were placed on ice, and a 10-l aliquot was diluted into 40 l of ice-cold Buffer C (50 mM Tris-Cl, pH 7.5, 75 mM NaCl, 5 mM MgCl 2 , 25 mM NaF, 5 mM EGTA, 100 M NaVO 4, and 1 mM DTT). 10 l of these diluted samples were taken into a second incubation with 5 l of MBP (16 g) and 10 l of an ATP mix containing 0.5 mM ATP, 50 mM MgCl 2 , [␥-32 P]ATP (2,400 cpm/pmol). The MBP kinase reaction was performed in duplicate. After 10 min the reaction was stopped by adding 6 l of 5ϫ SDS-PAGE sample buffer, and the reaction products were resolved on 15% SDS-PAGE gels. The radioactivity incorporated into MBP was measured by phosphorimaging after spotting the gels with a known amount of radioactive [␥-32 P]ATP. Background counts due to any P100associated MEK and ERK were estimated from the control tubes and subtracted from the assay counts (Ͻ5% total activity).
Raf-1 Immunoprecipitation Kinase Assays-To measure the specific activity of FLAGRaf, P100 fractions were normalized for FLAGRaf content, adjusted to 1% Nonidet P-40, sonicated for 90 s at 4°C, incubated on ice for 10 min, and microcentrifuged for 5 min, and the soluble membrane extract was diluted to 400 l with Buffer B (50 mM Tris-Cl, pH 7.5, 75 mM NaCl, 5 mM MgCl 2 , 25 mM NaF, 5 mM EGTA, 100 M NaVO 4, 1%Nonidet P-40, and 1 mM DTT). The samples were rotated with 10 l of anti-FLAG-Sepharose beads (Kodak) for 2 h at 4°C and washed six times in Buffer B. The immunoprecipitates were split into two aliquots and incubated with both MEK and ERK or with ERK alone, and the kinase assay was completed as described above. The beads were then collected, washed, and immunoblotted to verify the amount of Raf-1 present in the assay.
Immunofluoresence Assays-BHK cells were cultured at 37°C (5% CO 2) in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) bovine calf serum and 100 units/ml of penicillin and streptomycin. Cells were plated onto glass coverslips at 60% confluence and 4 h later were transfected using lipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions, with 1.6 g of EXV expression plasmid for Ha-RasG12V and Ha-Ras⌬G12V. After an overnight incubation, cells were fixed with 4% paraformaldehyde, permeabilized in 0.2% Triton X-100, and blocked with 3% bovine serum albumin in PBS. The cells were then incubated in undiluted Y13-238 hybridoma supernatant (anti-Ras). After extensive washing in PBS, cells were incubated in 30 g/ml fluorescein isothiocyanate-conjugated anti-rat secondary antibody (Pierce). Coverslips were washed and mounted in moviol. Fluorescence images were taken in a Bio-Rad MRC-600 Zeiss microscope using a 63ϫ magnification lens, a BHS filter, and blue light exciting at 488 Ϯ 5 nm with correction for emissions 515 nm and longer. Kalman averaging of eight scans was used to produce the final images.

RESULTS AND DISCUSSION
Ki-Ras Is a More Potent Activator of Raf-1 Than Ha-Ras-We first investigated whether Ha-Ras and Ki-Ras are equipotent activators of Raf-1. These two Ras isoforms were chosen for study because of their very different membrane localization signals. COS cells were transiently transfected with varying amounts of Ha-RasG12V and Ki-RasG12V plasmids, and the cells were fractionated into cytosol (S100) and membranes (P100). The amount of Ras expressed in each membrane fraction (Fig. 1A) was determined by quantitative immunoblotting A, membrane fractions were prepared from COS cells transfected with increasing amounts of Ha-RasG12V or Ki-RasG12V expression plasmids. These fractions were normalized for protein content, immunoblotted for Ras (upper panel) and Raf-1 activity measured in a coupled MEK/ERK activation assay (lower panel). B, MBP phosphorylation, which is the readout from the coupled Raf-1 activation assay, was measured by phosphorimaging and plotted against Ras expression (ng/20 g P100 protein) measured by quantitative Western blotting. The graph shows data pooled from three independent experiments; the images shown in A are from one of these experiments. against known amounts of recombinant Ras. The Raf-1 activity present in 20 g of cell membrane was measured in a coupled MEK/ERK activation assay using MBP phosphorylation as readout (Fig. 1A). The results from three such experiments are summarized in Fig. 1B. The data show that Ki-Ras is a significantly more potent activator of Raf-1 than Ha-Ras because minimal overexpression of Ki-Ras to 5 ng/20 g P100 (which equals 0.025% total membrane protein) resulted in a 5-fold greater activation of endogenous Raf-1 than the same minimal overexpression of Ha-Ras (Fig. 1B). The mechanism of Raf-1 activation is complex, but it is clear that one important role of Ras is to recruit Raf-1 to the plasma membrane (16,17,27,28) where a series of events is initiated that ultimately leads to full Raf-1 activation. These events include tyrosine, serine, and threonine phosphorylation (29 -33) plus interactions with Ras (20,34,35), phospholipids (36, 37), 14-3-3 proteins and their associated proteins (38 -44), and possibly dimerization (45,46). We reasoned, therefore, that the more potent activation of Raf-1 by Ki-Ras could reflect differences in the relative ability of Ki-Ras and Ha-Ras to recruit Raf-1 to the plasma membrane, and/or differences in Ki-Ras and Ha-Ras catalyzed membrane activation events.
To discriminate between these two possibilities COS cells were co-transfected with epitope-tagged FLAGRaf and varying amounts of Ha-RasG12V or Ki-RasG12V plasmids. S100 and P100 fractions were prepared from these transfectants and immunoblotted for Ras and FLAGRaf. The immunoblots were quantitated by phosphorimaging, and the data were used to calculate the fraction of FLAGRaf recruited to the membrane by increasing amounts of co-expressed Ras. The analysis in Fig.  2A shows that Ki-Ras recruits Raf-1 more efficiently than Ha-Ras, because recruitment of 50% of FLAG-Raf to the plasma membrane required expression of 20 ng (per 20 g of membrane) of Ki-Ras compared with 130 ng (per 20 g of membrane) of Ha-Ras ( Fig. 2A). We next determined whether Raf-1 recruited to the membrane fraction by Ki-and Ha-Ras differed in specific activity. Membrane fractions from the COS cells coexpressing FLAGRaf and Ki-or Ha-Ras were normalized for FLAGRaf content and immunoprecipitated using anti-FLAG-Sepharose. Raf-1 activity was then measured in the anti-FLAG immunoprecipitates. Significantly, Raf-1 recruited to the membrane by Ki-Ras was found to have a 4-fold higher specific activity than Raf-1 recruited to the membrane by Ha-Ras (Fig.  2B). We conclude that Ki-Ras is a more potent in vivo activator of Raf-1 because Ki-Ras recruits Raf-1 to the plasma membrane more efficiently than Ha-Ras and also because Ki-Ras more efficiently catalyzes Raf-1 activation at the cell membrane.
Influence of the Ras HVR on Raf-1 Membrane Recruitment-The polylysine domain of Ki-Ras (residues 175-180) interacts with negatively charged phospholipid head groups on the inner surface of the plasma membrane (15,47), residues C-terminal of residue 175 in Ki-Ras will therefore be tightly associated with the membrane. In contrast, the N-terminal limit of membrane tethering of Ha-Ras is the palmitoylation site at cysteine 181 (Fig. 3A). In consequence, the N-terminal conserved domains of Ras (1-166) are likely positioned closer to the plasma membrane in Ki-Ras than in Ha-Ras, i.e. by 7 Ki-Ras HVR residues (167-174) versus 13 Ha-Ras HVR residues (167-180). We therefore examined whether shortening the HVR of Ha-Ras by 6 amino acids would improve the ability of Ha-Ras to recruit and activate Raf. Ha-Ras⌬ was constructed by deleting amino acids 167-173, leaving intact all of the sequences required for plasma membrane localization (Fig. 3A). As expected, when expressed in BHK cells Ha-Ras⌬ was found to localize to the plasma membrane to the same extent as full-length Ha-Ras (Fig. 3B). The ability of Ha-Ras⌬ to recruit and activate Raf-1 was then compared with Ki-Ras and Ha-Ras. COS cells were co-transfected with FLAGRaf and varying amounts of Ha-Ras⌬G12V plasmid. Fig. 2A shows that shortening the HVR of Ha-Ras by 6 amino acids significantly improved its ability to recruit Raf-1 because the recruitment of 50% of expressed FLAGRaf required 35 ng (per 20 g of membrane) of Ha-Ras⌬ compared with 135 ng (per 20 g of membrane) of full-length Ha-Ras. The recruitment of Raf-1 to the plasma membrane by Ha-Ras⌬ therefore compares favorably with Ki-Ras ( Fig. 2A). However, the activation of membrane-recruited Raf-1 was not improved by shortening the Ha-Ras HVR because the specific activities of FLAGRaf recruited by Ha-Ras and Ha-Ras⌬ were not significantly different (Fig. 2B). Finally, we compared the membrane recruitment of endogenous Raf-1 by Ha-Ras⌬, Ha-FIG. 2. Ki-Ras recruits Raf-1 to the plasma membrane more efficiently than Ha-Ras and also activates membrane-recruited Raf-1 to higher specific activity than Ha-Ras. A, membrane and cytosolic fractions were prepared from COS cells co-transfected with a fixed amount of FLAGRaf and increasing amounts of Ha-RasG12V, Ki-RasG12V, or Ha-Ras⌬G12V expression plasmids. (Ha-Ras⌬G12V is described in Fig. 3.) The proportion of FLAGRaf recruited to the membrane fraction was determined by quantitative FLAG immunoblotting. The percentage of FLAGRaf recruited to the P100 fraction has been plotted against the Ras content of the membrane fraction (ng/20 g P100 protein) measured by quantitative Ras immunoblotting. B, equal amounts of FLAGRaf were immunoprecipitated (IP) from membrane fractions of COS cells coexpressing FLAGRaf and two different amounts of Ha-RasG12V, Ki-RasG12V, or Ha-Ras⌬G12V (in the range of 10 -50 ng/20 g P100 protein). The FLAG immunoprecipitates were assayed for Raf-1 activity and then immunoblotted for Raf-1 to verify normalization. The figure is of a representative experiment and shows that Ki-Ras activates Raf-1 to a higher specific activity than Ha-Ras or Ha-Ras⌬. WB, Western blot. C, because there was no increase in Raf-1-specific activity with increasing levels of Ras expression over the range of 10 -50 ng used for these experiments, data were pooled to calculate mean Raf-1 specific activity Ϯ S.E. (n ϭ 9).
Ras, and Ki-Ras. Fig. 3C shows that, as expected, low level of expression of Ki-Ras or Ha-Ras⌬ was sufficient to fully recruit all endogenous Raf-1 from the cytosol to the membrane, whereas a higher level of expression of Ha-Ras was required.
Together these experiments show that the different mechanisms of membrane attachment and the structure of the Ras HVR critically affect the in vivo interactions of Ras with one of its major effectors, Raf-1. The simple spacing of the Ras Nterminal domains from the plasma membrane may account for the different abilities of Ha-and Ki-Ras to recruit Raf-1 to the membrane. But there are two alternative explanations for the more potent activation of membrane-recruited Raf-1 by Ki-Ras: either Ki-Ras specific sequences in the HVR or elsewhere are required for maximal Raf-1 activation or by virtue of their membrane targeting signals Ha-and Ki-Ras recruit Raf-1 to distinct subdomains within the plasma membrane that contain different concentrations of Raf-1 activators.
Ha-Ras Is a More Potent Activator of Phosphoinositide 3-Kinase Than Ki-Ras-We next examined the ability of Ki-Ras and Ha-Ras to activate a second Ras effector, PI3-K. Anti-p85 antibodies were used to immunoprecipitate endogenous PI3-K from COS cells expressing increasing amounts of Ki-RasG12V and Ha-RasG12V. PI3-K activity was then measured in an in vitro lipid kinase assay using phosphatidylinositol as substrate. Fig. 4A shows the result of a representative experiment and Fig. 4B shows data pooled from four independent experiments. These data clearly show that Ha-Ras is a considerably more potent activator of PI3-K than Ki-Ras. Fig. 4 also shows that shortening the Ha-Ras HVR does not significantly affect the ability of Ha-Ras to activate PI3-K, because the dose response curves of Ha-Ras and Ha-Ras⌬ are similar. The more potent activation of PI3-K by Ha-Ras can be explained in two Ha-Ras compared with 9 HVR residues in Ki-Ras between the conserved Ras sequence  and the N-terminal limit of membrane attachment. The Ha-Ras cDNA sequences encoding residues His 166 though Leu 171 were deleted using oligonucleotide directed mutagenesis to generate Ha-Ras⌬, which therefore has 9 HVR residues between the conserved Ras sequence  and the N-terminal limit of membrane attachment. B, BHK cells were lipofected with expression plasmids for Ha-Ras⌬ and Ha-Ras (control) and Ras protein visualized by indirect immunofluoresence in a confocal microscope. Because the membrane targeting motifs are intact, Ha-Ras⌬ localizes normally to the plasma membrane. C, cytosol and membrane fractions of COS cells expressing increasing amounts of Ha-RasG12V, Ki-RasG12V, or Ha-Ras⌬G12V were normalized for protein content and immunoblotted for Ras and endogenous Raf-1. The figure shows that even a very low level of Ki-Ras expression is sufficient to fully recruit all endogenous Raf-1 to the membrane. Thus, when taken together with the data in Fig. 2, the increase in endogenous Raf-1 activation, shown in Fig. 1B, that is stimulated by Ki-Ras expressed at levels greater than 5 ng (per 20 g of P100 protein) may be largely mediated through an increase in endogenous Raf-1-specific activity. In contrast, the increase in endogenous Raf-1 activation stimulated by Ha-Ras expressed at levels greater than 5 ng (per 20 g of P100 protein) may result from both increased Raf-1 recruitment and increased Raf-1 specific activity.

FIG. 4. Ha-Ras is a more potent activator of PI3-K than Ki-Ras.
A, whole cell lysates (1 mg of total protein) prepared from cells expressing increasing amounts of Ha-RasG12V, Ki-RasG12V, or Ha-Ras⌬G12V were immunoprecipitated using anti-p85 antibodies. PI3-K activity associated with the immunoprecipitates was measured in an in vitro lipid kinase assay using phosphatidylinositol as substrate. The phosphorylated products of the reaction were resolved by TLC and visualized by phoshorimaging (upper panel). The arrow indicates radioactive PIP product identified on the basis of disappearance in cells treated with wortmannin (data not shown) and by co-migration with an unlabeled PIP standard, in turn visualized by iodine staining. The last lane on the TLC plate represents the PI3-K activity present in serum starved cells transfected with empty vector (control). Immunoprecipitates that are duplicates of those used in the kinase assay were immunoblotted for p85 to confirm that equivalent amounts of PI3-K were being captured from the cell lysates (middle panel). An aliquot of cells from each transfection was lysed in Triton-X114. The detergent partitioning protein fraction, containing processed, membrane localized Ras, was concentrated by acetone precipitation and immunoblotted for Ras (lower panel). B, after phosphorimaging of the TLC plates, Ras-stimulated PI3-K activity was calculated by subtracting the activity present in the control from the activity present in the Ras-transfected samples. The amount of biologically active Ras present in each lysate was determined by quantitative Western blotting of the Triton-X114 detergent partitioning fraction. Data are from four independent experiments, including that shown in A.
ways: efficient activation may require Ha-Ras specific sequences, located in the HVR or elsewhere, or Ha-Ras may recruit PI-3K to a distinct subdomain of the plasma membrane where activation proceeds more efficiently than in the subdomain to which PI3-K is recruited by Ki-Ras.
Concluding Remarks-Although all Ras isoforms qualitatively activate the same effector pathways, we have shown here that in vivo there are marked quantitative differences in the activation of c-Raf-1 and PI3-K by Ha-Ras and Ki-Ras. It seems probable that quantitative differences will extend to the activation of other Ras effector pathways, and therefore, activation of each Ras isoform can have distinct biochemical consequences for the cell. These different effector activation profiles may also account for the selective activation of different Ras isoforms in specific human tumors. We propose that the selection pressure for which Ras isoform is activated is determined by the background level of activity in all signaling pathways and that given the nature of multistep oncogenesis, this background activity will vary between target cells. In which case, activation of Ha-Ras may be favored in tumor cells that already have increased MAPK activity but low levels of PI3-K activity, whereas in tumor cells with constitutively elevated PI3-K activity, activation of Ki-Ras may be favored because it will result in coincident robust activation of the Raf/MAPK pathway. A recent study of human colorectal tumors adds significant weight to this hypothesis. Some 86% of human colorectal tumors tested were found to have significantly higher levels of PI3-K activity than normal colonic mucosa sampled from the same patient (48). Moreover, the level of PI3-K activity in these tumors was not further elevated in the presence of a Ki-ras mutation (48). This study is consistent with our finding that Ki-Ras is not a potent in vivo activator of PI3-K and, importantly, when taken together with the data presented here, rationalizes the preferential activation of Ki-Ras in colon cancer. In summary, our study emphasizes the importance of the mechanisms of plasma membrane attachment and the significance of the RasHVR in influencing the interaction of Ras with its effector proteins in vivo.