The dominant negative effects of H-Ras harboring a Gly to Ala mutation at position 60.

v-H-Ras harboring the Gly-60 to Ala mutation (G60A) lacks the ability to induce germinal vesicle breakdown in Xenopus oocytes. Moreover, this mutant is capable of inhibiting the activity of v-H-Ras to induce oocyte germinal vesicle breakdown when co-injected. The duration and the extent of inhibition depends on the molar ratio of v-H-Ras(G60A) to v-H-Ras. The inhibition is not due to a general toxicity of v-H-Ras(G60A) to oocytes because oocytes injected with v-H-Ras(G60A) can be readily induced to mature by other mitogenic agents, such as insulin, insulin-like growth factor 1, insulin-like growth factor 2, and phosphatidylcholine-specific phospholipase C. The dominant negative effect of v-H-Ras(G60A) requires proper membrane attachment of v-H-Ras(G60A). By using a competition assay, it was concluded that the dominant negative phenotype of v-H-Ras(G60A) resulted from sequestering H-Ras downstream effector(s). Raf-1 was identified as one of the sequestered targets.

Gly-60 of p21 ras is part of the conserved DXXG motif that is shared by members of the regulatory GTPase family (1,2). This sequence forms a sharply bent loop (loop 4) connecting ␤-strand 3 and ␣-helix 2 of H-Ras (20,21). GTP binding induces a conformational change primarily in two regions of H-Ras termed the switch I (residues 30 -38, effector domain) domain and the switch II (residue 60 -76) domain (22,23). The conformational change in the switch II domain clearly requires Gly-60 because significant structural alterations, such as the reorientation of the Gly-60 amide group and a partial collapse of helix 2, are found to be at or near this residue (20,21,23). Glycine is uniquely suited for this role because its range of and dihedral angles is broader than any other amino acid permitting flexible domain movements (24). Although this glycine is not conserved in all regulatory GTPases (25,26), its importance in the function of heterotrimeric Gs␣ protein, EF-Tu, and H-Ras is well documented (27)(28)(29)(30)(31). The G60A mutation in H-Ras 1 and corresponding mutations in Gs␣ (mutation G226A) and EF-Tu (mutation G83A) (27)(28)(29)(30) all perturb the GTP-induced conformational change in the switch II domain. In addition, the G60A mutation completely eliminates the biological activity of H-Ras without significantly affecting its binding to guanine nucleotides (31). It also drastically reduces RasGAP and NF1-induced GTPase activity of H-Ras but not the binding of H-Ras to them (32). The binding of H-Ras to Raf-1 was moderately attenuated by the G60A mutation (31); however, the binding of H-Ras to another downstream target, RalGDS (13,14,33), decreased by at least an order of magnitude more than Raf-1 binding (32). These results indicate that the switch II domain may participate, with varying degrees, in the interaction of H-Ras with downstream targets. In this study, we present evidence to show that H-Ras harboring the G60A mutation is a dominant negative mutant that inhibits the function of H-Ras by sequestering H-Ras downstream targets.

Mutant and Clone Construction-
The H-Ras G60A mutants were prepared as described (31,32). The double mutation v-H-Ras-(G60A,C186S) was produced by introducing the C186S mutation into the v-H-Ras(G60A) template by polymerase chain reaction as described (32). The polymerase chain reaction fragment was then digested with restriction enzymes BamHI and SalI and cloned into the BamHI-SalI site of plasmid pA-H-Ras (34). GST-Raf-1-N275 and GST-RalGDS were prepared as described previously (31,32).
Protein Preparation-All non-fusion Ras clones were expressed in the bacterial strain BL21(DE3) (35) and purified under non-denaturing conditions in the presence of GDP as described (34). The GST fusion proteins were purified using glutathione-Sepharose resin as described (31). GTP and GDP-bound H-Ras were prepared by guanine-nucleotide exchange in the presence of excess nucleotides as described (34). H-Ras proteins were freed of excess nucleotides before use by passing them through a Sephadex G50 column.
Oocyte Experiments-Stage VI oocytes were prepared from gonadotropin-primed adult female frogs by manual dissection as described (36). Barth's medium (37) was used throughout the experiment to maintain oocytes. Oocytes obtained from each frog were pre-tested before use by injecting 20 -30 oocytes with buffer and 46 ng of v-H-Ras. If more than 20% of the injected oocytes appeared in poor health, the entire batch of oocytes was not used for further experiments. For experiments, injections were always performed on a minimum of 10 -15 oocytes as a group by injecting 46 nl of the test materials into the vegetal hemisphere of oocytes. The injected oocytes were subsequently incubated at 19°C, and GVBD 2 was scored for up to 3 days by the appearance of a white spot in the animal hemisphere. If necessary, GVBD was further confirmed by dissection after fixing oocytes in 5% trichloroacetic acid. The data presented in the figures and tables are the average of at least three independent experiments from three individual frogs. Each experiment always included a set of oocytes injected with v-H-Ras and the combination of v-H-Ras and v-H-Ras(G60A) as the controls. GDP-bound H-Ras was used throughout the experiments unless stated otherwise. It should be pointed out that the v-H-Ras(G60A) dominant negative phenotype is greatly attenuated when nutrient-enriched DNOM medium (38) is used to incubate the oocytes instead of the Barth's medium. This enriched medium also significantly increased the rate of GVBD in oocytes injected with c-H-Ras or c-H-Ras(G60A). To simplify the process of characterizing the G60A mutant, Barth's medium was used throughout the study. In addition, we observed a seasonal variation in the effectiveness of the G60A dominant negative inhibition; therefore, all the experiments were performed between April and October, a period that the G60A mutant has the greatest effects. Bacillus cereus PC-PLC (grade I) and PI-PLC were obtained from Boehringer Mannheim. Recombinant human IGF-1 and IGF-II were purchased from Life Technologies, Inc.
Raf-1 Binding and Kinase Assay-H-Ras and Raf-1 complex formation was examined by a direct binding assay exactly as described (31). Raf-1 kinase activity was analyzed similarly as described by Muslin et al. (39). Oocytes were injected as described in the GVBD assay except that the injected oocytes were cultured for 4 h in Barth's medium at 19°C before analysis. Oocytes were homogenized in 0.5 ml of oocyte extraction buffer containing 20 mM Hepes, pH 7.2, 0.25 M sucrose, 0.1 M NaCl, 2.5 mM MgCl 2 , and 1 g/ml each of pepstatin, chymostatin, and leupeptin as described (40). The resulting lysate was centrifuged at 100,000 ϫ g for 20 min at 4°C to remove cell debris. The supernatant was then incubated with 20 l of agarose-conjugated anti-Raf-1 antibody (C-12, Santa Cruz Biotechnologies) overnight at 4°C to precipitate Raf-1. The agarose beads were washed four times with oocyte extraction buffer and twice with 0.5 M LiCl before assaying for Raf-1 kinase activity. To measure Raf-1 kinase activity, the Raf-1 immunoprecipitates were resuspended in 38 l of Raf-1 kinase reaction buffer containing 25 mM Tris-HCl, pH 7.5, 10 mM MnCl 2 , 10 M ATP, 1 mM dithiothreitol, 25 mM ␤-glycerophosphate, 10 Ci of [␥-32 P]ATP (specific activity, 7,000 Ci/mmol), and 1 l of 4 mM Syntide II peptide (Life Technologies, Inc.). The reaction was allowed to proceed at room temperature for 20 min. The reaction mixture was then spotted onto P81 phosphocellulose paper (Whatman), washed four times with 50 ml of 1.5% phosphoric acid and twice with 50 ml of 95% ethanol, and then air dried. The radioactivity remaining on the P81 paper was determined by liquid scintillation counting. Control experiments performed without the addition of oocyte lysate were used for background subtraction. Kinase activity is presented as the fold stimulation using oocytes injected with buffer as the base line (ϭ1).

The G60A Mutant Is a Dominant Negative Mutant against v-H-Ras-Introducing viral or mutagenic forms of p21 ras into
Xenopus oocytes readily induces GVBD (41). Previously, we showed that this activity in v-H-Ras was abolished by the G60A mutation (31). Consistent with the GVBD induction result, v-H-Ras(G60A) was also shown to lack transforming activity (31). However, v-H-Ras(G60A) is not biologically inert. When v-H-Ras(G60A) was injected together with v-H-Ras, it inhibited v-H-Ras-induced GVBD (Fig. 1A). In the range of protein concentrations used for injection (23-92 ng per oocyte), the duration and extent of inhibition depended on the molar ratio of v-H-Ras(G60A) to v-H-Ras. At a molar ratio of 0.5, v-H-Ras(G60A) partially inhibited v-H-Ras activity, while at a mo-lar ratio of 2, v-H-Ras(G60A) completely inhibited v-H-Ras for the entire 3-day incubation period (data not shown). This phenomenon was observed with both constant v-H-Ras (Fig. 1A) and v-H-Ras(G60A) (Fig. 1B). c-H-Ras harboring the same G60A mutation also displayed a similar but less potent dominant negative effect against v-H-Ras (Fig. 2). In contrast, c-H-Ras, a Ras form that does not promote GVBD, did not exert a dominant negative effect (Fig. 2). This observation shows that a lack of GVBD induction activity is not sufficient to produce dominant negative phenotype.

v-H-Ras(G60A) Disrupts the Interaction of H-Ras with Downstream
Targets-Subsequently, we investigated the possibility that v-H-Ras(G60A) may interfere with the interaction of v-H-Ras with downstream targets. These experiments were performed by supplementing the injection mixture with the suspected sequestered element as a binding competitor. Three different H-Ras downstream targets, RasGAP, Raf-1, and Ral-GDS, were tested using this approach. Fig. 4 shows that the human full-length RasGAP failed to reverse the effect of v-H-Ras(G60A) under all of the conditions tested (molar ratio Ras-GAP to v-H-Ras(G60A) up to 0.5). To exclude the possibility that the RasGAP preparation was biologically inactive, a control experiment using RAS T was performed. RAS T is a truncated yeast RAS1 gene product that harbors an additional Q61L mutation (47). RAS T exhibits an enhanced affinity for GAP and exerts a dominant negative effect against v-H-Ras that can be relieved by RasGAP (47). Fig. 4 shows that incorporating RasGAP in the injection mixture readily eliminated the RAS T dominant negative phenotype. This result indicates that the RasGAP preparation was biologically active; therefore, it excluded RasGAP as the potential cellular factor sequestered by v-H-Ras(G60A).
Subsequently, the ability of Raf-1 to compete against v-H-Ras(G60A) was examined. A GST-Raf-1 fusion protein containing the N-terminal first 275 residues of human Raf-1 (Raf-1-N275) was used in the study. This construct binds H-Ras⅐GTP in vitro (31) but does not induce oocyte GVBD because it lacks the Raf-1 catalytic domain (Fig. 5A). In contrast to RasGAP, the N-terminal fragment of Raf-1 protein was highly effective in suppressing the dominant negative effect of v-H-Ras(G60A). Fig. 5 shows that v-H-Ras(G60A) can be completely suppressed with approximately 1 GST-Raf-1-N275 per 27 v-H-Ras(G60A) molecules. In addition, the N-terminal 150 residues of Raf-1 were also able to suppress v-H-Ras(G60A) as efficiently as Raf-1-N275 fragment (data not shown). As a control, GST by itself has no v-H-Ras(G60A) suppression activity (data not shown). The ability of Raf-1 to suppress v-H-Ras(G60A) indicates that Raf-1 is either the direct downstream effector of v-H-Ras in the GVBD induction pathway or is capable of competing against the real H-Ras cellular effector for binding to v-H-Ras(G60A). The suppression of v-H-Ras(G60A) required nonstoichiomtric amounts of Raf-1-N275 to v-H-Ras(G60A); this observation suggests that only a fraction of the total cellular downstream effector (or Raf-1) is needed to propagate the H-Ras signal. The ability of Raf-1 to suppress v-H-Ras diminished when the molar ratio of Raf-1 to Ras approached unity (Fig. 5A). This phenomenon is apparently due to a direct inhibition of v-H-Ras by Raf-1 fragment at high concentrations (Fig. 5B); these concentrations are far greater than that required for suppressing v-H-Ras(G60A). The inhibition of viral or oncogenic Ras by Raf-1 N-terminal fragment has been reported by others (48,49).
We also tested RalGDS in the co-injection competition assay. The RalGDS fragment used here was a GST fusion protein containing the C-terminal 127 residues of the mouse RalGDS RBD (13,14,33). This fragment binds H-Ras⅐GTP with an affinity resembling that of GST-Raf-1-N275 (32). RalGDS-C127 fragment did not promote oocyte GVBD on its own (Fig. 6). Like GST-Raf-1-N275, introducing RalGDS-C127 into the injection mixture also abolished the dominant negative effect of v-H-Ras(G60A). Again, this result demonstrates that the site of v-H-Ras(G60A) inhibition is located downstream of H-Ras. However, 7-fold higher GST-RalGDS-C127 was required to suppress v-H-Ras(G60A) than was necessary using GST-Raf-1-N275 (Figs. 5 and 6). Furthermore, unlike the Raf-1-N275 fragment, GST-RalGDS-C127 at its optimum concentration (about one RalGDS for four Ras molecules) was not able to fully suppress the dominant negative effect of v-H-Ras(G60A). GST-RalGDS-C127 can also directly inhibit v-H-Ras-induced GVBD at high concentrations (data not shown). This may be why GST-RalGDS beyond a certain level causes an apparent reduction in v-H-Ras(G60A) suppression efficiency (Fig. 6); this result is similar to that observed in the competition assay with Raf-1-N275 (Fig. 5).

The Dominant Negative Effect of v-H-Ras(G60A) Requires Proper Membrane Attachment of v-H-Ras(G60A)-Post-trans-
lational modification of Ras near its C terminus is required for its membrane localization and biological activity (50). To determine whether membrane attachment is essential for the dominant negative phenotype of v-H-Ras(G60A), we introduced an intragenic C186S mutation into v-H-Ras(G60A); this mutation blocks Ras membrane attachment. Although the C186S mutation did not alter the binding of H-Ras to Raf-1-N275 in vitro (Table III), it completely abolished the dominant negative effect of v-H-Ras(G60A) ( Table III). This experiment indicates that v-H-Ras(G60A) requires proper membrane attachment to sequester the downstream effector v-H-Ras.

The Inhibition of v-H-Ras Induced Raf-1 Activity by v-H-Ras(G60A)-The direct effects of v-H-Ras(G60A) on v-H-Ras-
induced Raf-1 kinase activity was analyzed. As was observed in the GVBD induction studies, introducing v-H-Ras into oocytes readily induced Raf-1 kinase activity; this activity was greatly diminished by the G60A mutation. However, unlike GVBD induction, the G60A mutant still exhibited detectable levels of Raf-1 kinase induction activity (Fig. 7). In accord with its ability to inhibit v-H-Ras-induced GVBD, the presence of v-H-Ras(G60A) drastically reduced the ability of v-H-Ras to induce Raf-1 kinase activity. This reduction can be fully restored by co-injecting low levels of GST-Raf-1-N275 or GST-RalGDS-C127 fragment (Fig. 7). As controls, we found that Raf-1-N275 or RalGDS-C127 fragments alone did not induce Raf-1 kinase (Fig. 7). We also found that both c-H-Ras and v-H-Ras-(G60A,C186S) were unable to inhibit v-H-Ras-induced Raf-1 kinase (data not shown). In summary, our results identify Raf-1 as a target inhibited by v-H-Ras(G60A). DISCUSSION We have described a new H-Ras dominant negative mutant, v-H-Ras(G60A), that inhibits the activity of v-H-Ras in an oocyte GVBD induction assay. v-H-Ras(G60A) was found to specifically inhibit v-H-Ras but not other mitogenic agents, such as insulin and IGF. The mechanism of inhibition is unlikely to be due to perturbation of guanine-nucleotide exchange factor because v-H-Ras(G60A) was equally effective against either v-H-Ras⅐GDP or v-H-Ras⅐GTP. We then searched for v-H-Ras inhibition site among H-Ras downstream targets using a competition assay. Both the N-terminal 275 residues Raf-1 and the C-terminal 127 residues RalGDS were found to suppress the dominant negative effect of v-H-Ras(G60A). These findings indicate that v-H-Ras(G60A) inhibits H-Ras downstream target(s). Unfortunately, H-Ras interacts with multiple downstream targets, and more than one target (Raf-1 and RalGDS) suppressed v-H-Ras(G60A); therefore, the true downstream target(s) of H-Ras could not be determined solely by the competition assay. Nevertheless, by combining the findings of GVBD and kinase assays, it is clear that Raf-1 is an inhibited target. This conclusion is in accord with numerous other studies (51,52). However, since our current results cannot rule out the possibility that v-H-Ras(G60A) may also inhibit other cellular factors involved in the H-Ras signaling pathways (19,53), it is not clear whether Raf-1 kinase inhibition alone can fully explain the dominant negative phenotype of v-H-Ras(G60A).
The straightforward interpretation for the Raf-1 kinase inhibition is that v-H-Ras(G60A) forms a stable but nonproduc-  The GVBD was scored 24 h after the injection using 46 ng each of v-H-Ras and the G60A mutant per oocyte. The numbers in parentheses, pooled from three independent experiments, indicated the number of matured and total oocytes, respectively. The Ras coprecipitation assay was performed as described (31).
NA, not applicable. tive complex with Raf-1; this eventually depletes the intracellular Raf-1 pool and blocks the activity of v-H-Ras. However, this interpretation is in direct contrast to the hypothesis that Raf-1 membrane recruitment is the sole contribution of H-Ras to Raf-1 activation (54,55). If the hypothesis were true, v-H-Ras(G60A) should be able to induce Raf-1 kinase and in turn activate subsequent mitogenic responses because it binds well to Raf-1. Apparently, membrane recruitment is a necessary but not a sufficient step for Raf-1 activation. Membrane recruitment of Raf-1 by H-Ras is mediated by the interaction between the effector domain (switch I) of H-Ras and Raf-1 residues 55-132 termed RBD (8 -11, 48, 56, 57). Since the switch I domain is not perturbed by the G60A mutation, the binding of H-Ras to Raf-1 RBD is basically intact in the G60A mutant (31,32,58). Recently, another Raf-1 RBD consisting of residues 130 -196, which is capable of independently binding Ras⅐GTP, was identified (49). Interestingly, in contrast to Raf-1 fragments containing the first RBD, the binding of the Raf-1 second RBD is severely impaired by the G60A mutation (58). These findings support the hypothesis that Ras induces Raf-1 through a multiple step reaction, which may involve an initial interaction between the first RBD with the switch I domain that recruits Raf-1 to membrane and a second interaction between the second RBD and the switch II domain that initiates Raf-1 activation. It appears that the G60A mutant sequesters Raf-1 by uncoupling these interaction events. v-H-Ras(G60A) cannot display its dominant negative phenotype without proper membrane attachment (Table III). This result not only indicates that membrane localization is important for sequestering Raf-1, it also implies that the second interaction between Ras and Raf-1 may not occur until Raf-1 is membrane bound. However, it is not known whether membrane attachment per se is sufficient or if additional factors are required to form stable v-H-Ras(G60A)⅐Raf-1 complexes.
In addition to forming inactive v-H-Ras(G60A)⅐Raf-1 complexes, it is clear that the competitive binding of v-H-Ras and v-H-Ras(G60A) to their downstream effectors also contributes to the dominant negative effect of v-H-Ras(G60A). This is why the extent of v-H-Ras(G60A) dominant negative effect is critically dependent upon the molar ratio of v-H-Ras(G60A) to v-H-Ras ( Figs. 1 and 2). The GTPase activity of c-H-Ras was drastically reduced by the G60A mutation; however, unlike v-H-Ras(G60A), c-H-Ras(G60A) still contained detectable GTPase activity (32). Therefore, c-H-Ras(G60A)⅐GTP must have a shorter half-life than v-H-Ras(G60A)⅐GTP. GTP hydrolysis promotes the dissociation of the sequestered downstream target from the sequestered complex; this may be why c-H-Ras(G60A) exhibits a weaker dominant negative phenotype than v-H-Ras(G60A).
Apparently, the ability to form a stable complex with v-H-Ras(G60A) is not a universal property of all H-Ras downstream targets. This phenomenon may simply reflect the difference in binding strength. Raf-1 and RalGDS are known to bind much more tightly to H-Ras than RasGAP (13,59,60); therefore, the amount of RasGAP required to compete against Raf-1 and subsequently relieve GVBD inhibition may not be attainable by co-injection. We have not tested other H-Ras downstream targets, such as phosphatidylinositol 3-kinase, MEKK, PKC, and Rin (15,17,19,61) in the competition assay; however, if any of these factors are able to bind H-Ras as tightly as Raf-1 or RalGDS, they should suppress v-H-Ras(G60A).
Although the G60A mutation reduces the binding of H-Ras to both Raf-1 and RalGDS (K d ϭ 0.236 M for the v-H-Ras(G60A)⅐GST-Raf-1-N275 complex and 2.05 M for the v-H-Ras(G60A)⅐GST-RalGDS-C127 complex as compared with 0.047 M and 0.037 M, respectively, for v-H-Ras⅐GST-Raf-1-N275 and v-H-Ras⅐GST-RalGDS-C127 complexes) (32), this deficiency can be readily compensated in injected oocytes. Assuming 500 nl for the volume of an oocyte, Ras at 46 ng, GST-Raf-1-N275 at 4.6 ng, and GST-RalGDS-C127 at 23 ng (Figs. 5 and 6) will have intracellular concentrations of 4.3, 0.16, and 1.1 M, respectively. Under these conditions, v-H-Ras(G60A) should bind rather well to Raf-1 or RalGDS fragment and reduce its ability to sequester the true downstream target. Since it only requires a small amount of Raf-1 and RalGDS fragment to suppress v-H-Ras(G60A), this result suggests that oocytes need only a small percentage of total cellular downstream targets (or Raf-1) for propagating H-Ras signal. The same mechanism also explains the finding that high amounts of Raf-1 or RalGDS fragments are required to compete directly against the activity of v-H-Ras (Fig. 5B). Otherwise, any amount of competing Raf-1 or RalGDS fragment would have caused measurable v-H-Ras inhibition. A similar observation was reported by others using Raf-1 RBD fragment and oncogenic N-Ras (48). 7-fold more GST-RalGDS-C127 is needed to achieve a comparable level of v-H-Ras(G60A) suppression than that of GST-Raf-1-N275 (Figs. 5 and 6). This phenomenon appears to be due to the difference in binding strength since RalGDS binding is more drastically reduced (by about 8-fold) by the G60A mutation than Raf-1 binding (32). Furthermore, the much reduced RalGDS binding by the G60A mutation argues against the possibility that RalGDS is a v-H-Ras(G60A) sequestered target unless oocytes contain very high endogenous RalGDS. The threshold appears to be around the equivalent of 23 ng of GST-RalGDS-C127 per oocyte since this is the concentration of RalGDS that can efficiently suppress v-H-Ras(G60A) (Fig. 6). Thus, RalGDS may only function as an opportunistic v-H-Ras(G60A) suppressor in our assay by virtue of its ability to titrate out v-H-Ras(G60A) at high concentrations.
The membrane attachment requirement for v-H-Ras(G60A) to exert its dominant negative phenotype of v-H-Ras(G60A) distinguishes it from the RAS T class of dominant negative mutants (47,62), which need cytosolic proteins. Intriguingly, we found that while Raf-1 was highly effective against v-H-Ras(G60A), it could not suppress Ras T . On the other hand, RasGAP suppressed Ras T but not v-H-Ras(G60A). These observations imply that H-Ras requires both RasGAP and Raf-1 to transmit its mitogenic response. We cannot reconcile this apparent paradox at the present time. Since anti-Ras monoclonal antibodies Y13-259 and 6B7 have been shown to block insulininduced oocyte GVBD (63,64), it was surprising to find v-H-Ras(G60A) had no effect against insulin (Table I). This result implies that insulin may employ signaling route(s) other than Raf-1, the known v-H-Ras(G60A) inhibition point, to promote GVBD. Several potential candidates exist, such as MEKK, REKS, and others (16,53,65,66) that have been shown to activate MEK and subsequent mitogen-activated protein kinase pathway. These MEK activators may represent possible insulin bypass points. It is also possible that v-H-Ras(G60A) inhibition is bypassed by other members of Raf family, such as A-and B-Raf (67). We are currently investigating these possibilities.
Ras harboring the G60A mutation is a potent inhibitor that perturbs the H-Ras signaling pathway by sequestering essential downstream effector(s). The properties of the G60A mutant will be useful for elucidating the mechanism of H-Ras interacting with its downstream effectors as well as the signal transduction pathway(s) associated with H-Ras. and Robert A. Weinberg for the mouse RalGDS clone. We also thank Drs. Robert B. Denman and David L. Miller for critical reading of this manuscript.