N-terminal Sequences Contained in the Src Homology 2 and 3 Domains of p120 GTPase-activating Protein Are Required for Full Catalytic Activity Toward Ras*

The p120 GTPase-activating protein (GAP) is a nega- tive regulator of Ras, which has a central role in signal transduction pathways that control cell proliferation. p120 GAP accelerates the conversion of activated Ras-GTP to its inactive form, Ras-GDP, thereby inhibiting mitogenic signaling. To examine potential contributions of p120 N-terminal sequences to regulation of its C-ter- minal catalytic domain, we constructed deletion mutants lacking defined regions, including the variable hy- drophobic region as well as the Src homology 2 (SH2) and 3 (SH3) domains. These mutant proteins were ex- pressed in infected Sf9 insect cells from recombinant baculoviruses and assayed in vitro for their ability to stimulate the intrinsic GTPase activity of purified Ras. While deletion of the variable hydrophobic region had no effect on p120 GAP activity, deletion of the entire SH2/SH3/SH2 region severely impaired catalytic activity toward Ras. Deletion of individual SH2 and SH3 domains within this region partially inhibited p120 GAP activity. Moreover, p120 N-terminal sequences en-hanced the Ras GTPase-stimulating GAP and human NF1 cDNAs were also made and expressed in baculovirus-infected Sf9 insect cells. The p120 catalytic domain was fused to the N-terminal sequences of human neurofibromin (NF1-GAP). A reciprocal construct was made where the neurofibromin catalytic domain (NF1 GRD) was fused to the N-terminal sequences of p120 GAP (GAP-NF1). An additional recombinant was constructed containing the isolated GRD of neurofibromin. PH , pleckstrin homology domain; Ca calcium-dependent lipid binding domain.

p21 Ras functions as a binary switch in mitogenic signal transduction pathways, and, as a key signaling molecule, its activation must be tightly regulated in order to control cell proliferation (1)(2)(3)(4)(5). Activated Ras is bound to GTP, while the GDP-bound form is inactive. Ras can hydrolyze bound GTP to GDP, resulting in self-inactivation; however, this intrinsic GTPase activity is low and requires an additional factor to stimulate the activity (6,7). Stimulation is provided by Rasspecific GTPase-activating proteins (Ras GAPs), 1 including p120 GAP and the product of the NF1 gene, neurofibromin. Both of these GAPs accelerate the conversion of activated Ras-GTP to its inactive GDP-bound form, thereby downregulating mitogenic signaling by Ras (7)(8)(9)(10)(11)(12).
While the catalytic domain of p120 GAP is contained in its C-terminal portion (13), the contribution of its N-terminal sequences to catalytic activity is not well defined. The N-terminal sequences harbor a hydrophobic region that varies among species, an SH3 domain flanked by two SH2 domains, as well as pleckstrin homology and calcium-dependent lipid binding domains (14 -17). Neurofibromin contains a central catalytic domain with high sequence similarity to the p120 GAP catalytic domain, but unlike p120 GAP, it has no identifiable domains outside of this region. Even though the structural differences between p120 GAP and neurofibromin are likely to reflect different cellular functions, if either protein fails to negatively regulate Ras, altered downstream signaling events will occur. Loss of neurofibromin contributes to neoplasia and can result in von Recklinghausen's neurofibromatosis (18 -21). In addition, point mutations in one of the SH2 domains of p120 GAP have been detected in basal cell carcinomas (22), consistent with a role for these mutations in neoplasia.
We have shown previously that the SH2 domains in p120 GAP mediate its interaction with tyrosine-phosphorylated p190, which is a GAP for the Rho/Rac family of proteins (23,24). Although p120 GAP SH2 and SH3 domains have been shown to mediate several different protein-protein interactions (25)(26)(27), there is also evidence suggesting that they may be involved in regulating GAP activity toward Ras. Previous studies showed that deletion of the entire N-terminal half of p120 GAP, including the SH2/SH3/SH2 region, leads to a significant loss in GAP activity, thus implicating N-terminal sequences as necessary for maximal p120 GAP catalytic activity (28). On the other hand, different studies have suggested a possible negative regulatory role for the p120 GAP N-terminal sequences (29). The specific domains in the p120 GAP N-terminal region that are involved in regulating catalytic activity, however, have not been defined.
To examine contributions of p120 GAP N-terminal sequences to GTPase-stimulating activity, we constructed deletion mutants lacking defined regions, including the variable N-terminal hydrophobic region and the SH2/SH3 domains. Analysis of these mutants showed that deletion of the entire SH2/SH3/SH2 region severely impaired p120 GAP catalytic activity toward Ras, while deletion of the individual SH2 and SH3 domains partially impaired GAP activity. Consistent with a positive role for the SH2/SH3/SH2 region, p120 N-terminal sequences enhanced the Ras GTPase-stimulating activity of the GAP-related domain (GRD) of neurofibromin. Conversely, a reciprocal domain swap with the neurofibromin N-terminal sequences did not restore activity of the p120 GAP catalytic domain. Our results suggest that sequences within the SH2/SH3/SH2 region of p120 GAP are necessary for maximal activity toward Ras and have important implications for the potential contribution of mutations in this region to oncogenesis.

MATERIALS AND METHODS
Baculovirus Recombinants-Construction of the following recombinant baculovirus vectors has been described previously (30): bGAP encoding full-length bovine p120 GAP and ⌬SH encoding p120 GAP with a deletion of amino acids 166 -518. Using polymerase chain reaction deletion mutagenesis strategies, the following bovine p120 GAP baculovirus recombinants were constructed: ⌬Nterm missing amino acids 25-164, ⌬NSH2 missing amino acids 177-268, ⌬SH3 missing amino acids 282-336, and ⌬CSH2 missing amino acids 342-435. The GAP-NF1 hybrid contains amino acids 1-691 of human p120 GAP fused to amino acids 1168 -1614 of human neurofibromin, while the NF1-GAP recombinant contains amino acids 1-1167 of neurofibromin fused to amino acids 671-1149 of p120 GAP. 2 The GRD construct contains amino acids 1125-1536 of human neurofibromin encompassing the catalytic domain of the protein.
Cell Culture-Spodoptera frugiperda (Sf9) insect cells (American Type Culture Collection) were cultured as described previously (30,31). For protein production, Sf9 cells were singly infected with recombinant baculovirus stocks using a multiplicity of infection of 10 for each virus.
Protein Expression-48 h postinfection, infected Sf9 cells were washed 3 times in cold phosphate-buffered saline and resuspended in sonication buffer (20 mM Tris-HCl, pH 7.5, 100 mM KCl, 1 mM MgCl 2 , 10% glycerol, 1 mM sodium orthovanadate, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride, 1 M leupeptin, 1 M antipain, and 0.1 M aprotinin). Cells were then lysed by sonication and clarified by centrifugation, and the soluble protein-containing supernatant was used in subsequent catalytic assays. Normalization of GAP protein levels was achieved by quantitative Western blotting analysis (23). Proteins were resolved on a 7.5% SDS gel, transferred to nitrocellulose, and probed with monoclonal anti-GAP, polyclonal anti-GAP, or polyclonal anti-GRD antibodies. Primary antibodies were detected by ECL (Amersham Corp.) using horseradish peroxidase-conjugated antimouse or anti-rabbit antibodies, or by 125 I-protein A (ICN) and subsequent quantification by PhosphorImager (Molecular Dynamics).
GAP Catalytic Assay-Purified bacterially-expressed human c-Ha-Ras (13) was loaded with [␥-32 P]GTP (30 Ci/mmol, DuPont NEN) in the presence of 1 mM EDTA, 1 mM dithiothreitol, 0.25 mg/ml bovine serum albumin, and 22 mM Tris-HCl, pH 7.5, at 37°C for 10 min as described previously (32,33). GTP-loaded Ras was then incubated with insect cell lysates containing baculoviral-expressed GAPs in the presence of 5 mM MgCl 2 , 13 mM Tris-HCl, pH 7.5, 0.1 mg/ml bovine serum albumin, 1 mM dithiothreitol, and 1 mM GTP at 37°C for the indicated times. Reactions were quenched on ice in 1 ml of ice-cold washing buffer (50 mM NaCl, 5 mM MgCl 2 , and 25 mM Tris-HCl, pH 7.5), and filtered through nitrocellulose filters (0.45 m, Schleicher and Schuell). Initial loading at zero time points was determined by adding GTP-loaded Ras to prequenched reactions. GAP activity was measured by monitoring the decrease in [␥-32 P]GTP-Ras bound to the filter by liquid scintillation spectrometry (32,33). For assays using full-length p120 GAP and the individual ⌬NSH2, ⌬SH3, and ⌬CSH2 deletion mutants, initial velocities of p120 GAP activities were determined over a range of concentrations of p21 Ras-GTP. Levels of p120 GAP used in these experiments were previously determined to be in the linear range of the assay. Initial velocities of GTPase reactions in the presence of full-length p120 or deletion mutants were fitted to the Michaelis-Menten equation using Sigmaplot Curvefit.
Antibodies-Monoclonal anti-p120 GAP 6F2 antibody was raised against the SH3 domain of human p120 GAP and affinity purified over a protein G column (34). Monoclonal anti-p120 GAP 7D1 antibody was raised against the calcium-dependent lipid binding domain of human p120 GAP (34). Anti-p120 GAP 677 rabbit polyclonal sera was raised against amino acids 988-1001 of bovine GAP (13), while polyclonal anti-GRD sera was raised against amino acids 1400 -1419 of the human neurofibromin GRD (35).

Deletion of p120 GAP SH2/SH3/SH2 Region Impairs its
Catalytic Activity Toward Ras-To examine potential contributions of the p120 N terminus to GAP activity, we created several p120 deletion mutants missing various regions and domains in the N-terminal portion of the protein. Mutants were constructed that lack the N-terminal variable hydrophobic region, the entire SH2/SH3/SH2 region, or individual SH2 and SH3 domains (Fig. 1). Full-length p120 GAP and deletion mutants were expressed in Sf9 insect cells infected with recombinant baculoviruses. Infected cells were disrupted by sonication, and expression levels of soluble recombinant proteins were normalized to full-length p120 GAP by quantitative Western blot analysis for use in subsequent catalytic assays.
GAP catalytic activity was measured by an in vitro filterbinding assay using purified, bacterially-expressed p21 c-Ha-Ras loaded with [␥-32 P]GTP as substrate. Background GAPlike activity in the assay was found to be negligible when the activity of nonrecombinant baculovirus-infected Sf9 whole cell lysates was compared with the activity of Ras alone (data not shown). Using subsaturating Ras conditions, we compared fulllength p120 GAP with two p120 recombinants, ⌬SH and FIG. 1. p120 GAP and neurofibromin recombinants expressed from baculoviruses. Full-length bovine p120 GAP cDNA was used to make several deletion mutants. In addition to wild-type p120 GAP, constructs lacking part of the N-terminal hydrophobic region, the entire SH2/SH3/SH2 region, or the individual SH2 and SH3 domains were expressed from baculovirus recombinants in infected Sf9 cells. Chimeras of human p120 GAP and human NF1 cDNAs were also made and expressed in baculovirus-infected Sf9 insect cells. The p120 catalytic domain was fused to the N-terminal sequences of human neurofibromin (NF1-GAP). A reciprocal construct was made where the neurofibromin catalytic domain (NF1 GRD) was fused to the N-terminal sequences of p120 GAP (GAP-NF1). An additional recombinant was constructed containing the isolated GRD of neurofibromin. PH, pleckstrin homology domain; Ca, calcium-dependent lipid binding domain.
⌬Nterm, and the GAP activity is shown as percent of GTP hydrolysis (Fig. 2). Deletion of the SH2/SH3/SH2 region in ⌬SH markedly impaired p120 GAP catalytic activity, resulting in a level of GTP hydrolysis comparable with that of nonrecombinant baculovirus-infected cell lysates alone ( Fig. 2A). In contrast, deletion of the N-terminal hydrophobic region in ⌬Nterm had no effect on p120 GAP activity (Fig. 2B), suggesting that this region is not essential for full catalytic activity toward Ras.
Each Domain within the SH2/SH3/SH2 Region of p120 GAP Positively Contributes to p120 GAP Catalytic Activity-To define contributions of the individual domains within the SH2/ SH3/SH2 region to p120 GAP activity, additional mutants with precise deletions of the SH2 and SH3 domains (Fig. 1) were assayed for catalytic activity. We performed more extensive assays designed to saturate GAP activity with increasing concentrations of Ras-GTP in order to detect subtler effects on p120 GAP activity that might potentially result from the individual domain deletions. Full-length p120 GAP and the deletion mutants were incubated with increasing concentrations of [␥-32 P]GTP-Ras ranging from 0 to 50 M, and the initial velocity of the GTPase reaction was determined at each concentration. Initial velocities were plotted as a function of Ras concentration and fitted to the Michaelis-Menten equation. Saturation of the GAP activity was not achieved under our experimental conditions, precluding accurate determination of the catalytic constants. However, since we normalized the GAP protein levels by quantitative Western blot analysis, we can readily detect relative differences in catalytic activity of deletion mutants as compared with full-length p120 GAP.
Consistent with results obtained under subsaturating Ras concentrations, the ⌬SH mutant was severely impaired in its activity toward Ras compared with full-length p120 GAP (Fig.  3A). We assayed the individual p120 domain deletion mutants and found that the ⌬NSH2, ⌬SH3, and ⌬CSH2 deletion mutants were all partially impaired to similar extents in their activity toward Ras (Fig. 3, B-D). While the activities of the individual p120 domain deletion mutants were higher than that of ⌬SH, their activities were reproducibly lower than that of full-length p120. Relative levels of GAP proteins used in each assay are shown as insets to the graphs. These results indicate that the presence of each domain within the SH2/SH3/SH2 region is necessary for p120 GAP to display full catalytic activity toward Ras.
Sequences within the N Terminus of p120 GAP Stimulate Neurofibromin GAP Activity Toward Ras-We examined the effect of p120 N-terminal sequences on the catalytic activity of another Ras GAP, the NF1 gene product, neurofibromin. Recombinant proteins containing the N-terminal half of p120 GAP fused to the neurofibromin GRD (GAP-NF1) or containing the isolated neurofibromin GRD were expressed in baculovirusinfected Sf9 cells (Fig. 1). Analysis of these mutants showed that the neurofibromin GRD was impaired in its GTPase-stimulating activity toward Ras but retained some activity above what was seen with the nonrecombinant vector control (Fig.  4B). When the N-terminal portion of p120 GAP was fused to the neurofibromin GRD, however, GAP activity was significantly enhanced.
To determine whether any sequence fused N-terminal to the p120 GAP catalytic domain enhances activity, a reciprocal recombinant (NF1-GAP) was analyzed that encodes the catalytic domain of p120 GAP downstream of the N-terminal portion of neurofibromin (Fig. 1). Significantly, NF1-GAP had only marginally more activity than the nonrecombinant baculovirus control for background (Fig. 4A). Unlike what was observed with the GAP-NF1 hybrid, the N-terminal sequences of neurofibromin did not have a positive effect upon p120 GAP activity. This finding is consistent with the suggestion that the enhanced activity of the p120 GAP compared with ⌬SH is due to FIG. 2. Deletion of the entire SH2/ SH3/SH2 region severely impairs p120 GAP catalytic activity toward Ras. Lysates of cells infected with baculoviruses encoding ⌬SH or ⌬Nterm were normalized for expression levels to fulllength p120 GAP and then assayed for their ability to stimulate Ras intrinsic GTPase activity. GAP activity is shown as % GTP hydrolysis with time using 80 nM [␥-32 P]GTP-loaded Ras (see "Materials and Methods"). As a negative control, lysates of Sf9 cells infected with nonrecombinant baculovirus (Baculo.) were used to detect any background GTPase activity. Plots from both ⌬SH (A) and ⌬Nterm (B) represent the means of three separate experiments with standard errors shown.
FIG . 3. The individual NSH2, SH3, and CSH2 domains of p120 GAP contribute to maximal GAP activity. A, clarified lysates from Sf9 cells infected with recombinant viruses encoding full-length p120 GAP and ⌬SH were resolved by SDS-polyacrylamide gel electrophoresis, and Western blots were probed with monoclonal anti-GAP antibody 7D1. Normalized proteins were subjected to GAP catalytic assays using increasing concentrations of purified c-Ha-Ras preloaded with [␥-32 P]GTP. The initial velocities of Ras-GTP hydrolysis as measured by pmol of phosphate released/min reflect GAP catalytic activity. Similar assays were performed on individual domain deletion mutants: B, ⌬NSH2; C, ⌬SH3; and D, ⌬CSH2. Recombinant protein levels were normalized by Western blot analysis to full-length p120 GAP (as shown in graph insets) with monoclonal anti-GAP antibodies 6F2 (B, D) or 7D1 (C) as probes. Results represent the means of two separate experiments, with standard errors shown. specific sequences within the p120 GAP N-terminal portion. Taken together with results from the SH2 and SH3 domain deletion mutants, our data demonstrate that the SH2/SH3/SH2 region of p120 GAP is required for full catalytic activity toward Ras and that the N-terminal sequences of p120 GAP stimulate neurofibromin GRD activity. DISCUSSION Because negative regulation of Ras activity by GAPs is one crucial mechanism the cell employs to control Ras, regulation of GAP activity is an important event for normal cell growth (8, 16, 36 -38). The N-terminal portion of p120 GAP has been implicated in a variety of GAP interactions and functions in the cell. The N-terminal hydrophobic region has recently been shown to contribute to interactions with Src-family members through its proline-rich motif (39). The SH2 domains are responsible for multiple protein-protein interactions, including interactions between GAP and receptor tyrosine kinases (40,41), nonreceptor tyrosine kinases (30,42), as well as the GAPassociated proteins, p62 and p190 (23,(43)(44)(45). Furthermore, the intact SH2/SH3/SH2 region is capable of uncoupling a heterotrimeric G protein from muscarinic receptors (29), while the p120 GAP SH3 domain alone blocks germinal vesicle breakdown in Xenopus oocytes (46) and inhibits carbachol-dependent NIH 3T3 transformation via muscarinic receptors (47,48). In addition to affecting biological functions within different cells, the entire N-terminal half of p120 GAP has been suggested to positively regulate GAP catalytic activity in one case, and possibly inhibit it in another (28,29). To address the question of the contribution of p120 GAP N-terminal sequences, especially its SH2/SH3/SH2 region, to p120 GAP catalytic activity toward Ras, we constructed several p120 GAP deletion mutants, expressed them in the baculovirus/Sf9 insect cell system, and assayed their ability to stimulate Ras intrinsic GTPase activity.
Consistent with earlier results involving deletion of the entire N terminus of p120 GAP (28), deletion of the SH2/SH3/SH2 domains significantly impaired p120 GAP activity toward Ras. The ⌬SH mutant did not detectably stimulate GTPase activity above that of the Ras intrinsic GTPase activity as reflected by the nonrecombinant baculovirus control. Other studies indicated that the isolated p120 GAP catalytic domain displayed reduced but detectable activity (10,12,13,28); this difference might be explained by different constructs or levels of protein used in the assays. By contrast, analysis of the ⌬Nterm mutant revealed that deletion of this region had no affect upon p120 GAP catalytic activity. These findings suggest a specific requirement for the SH2/SH3/SH2 region for maximal p120 GAP catalytic activity. We examined which domains within this SH2/SH3/SH2 region were responsible for the reduced activity of the ⌬SH GAP construct. p120 GAP deletion mutants lacking the individual N-terminal SH2 or SH3 or C-terminal SH2 domains were also impaired in their ability to stimulate Ras GTPase activity, albeit to a lesser extent. No single domain deletion impaired p120 GAP activity as much as deleting the entire SH2/SH3/SH2 region did, suggesting that the complete lack of activity of the ⌬SH mutant may be due to the cumulative effect of the individual domain deletions.
To explore further the stimulatory role of the p120 N-terminal sequences, we investigated whether they could affect the GAP activity of the NF1 protein, neurofibromin. When the N terminus of p120 GAP was fused with the GRD of neurofibromin (GAP-NF1), this construct exhibited much higher GAP catalytic activity than the neurofibromin GRD alone. In addition, fusion of p120 N-terminal sequences to the neurofibromin GRD stimulated activity above that of a near full-length neurofibromin recombinant containing the N-terminal neurofibromin sequences as well as the GRD. 3 Significantly, the reciprocal domain swap did not restore p120 GAP activity, as shown by analysis of the hybrid containing the N-terminal sequences of neurofibromin and the catalytic domain of p120 GAP (NF1-GAP). This finding, taken together with the analysis of p120 GAP deletion mutants, points to sequences within the p120 GAP SH2/SH3/SH2 domains as being necessary for p120 GAP to maximally stimulate GTPase activity of Ras.
While the mechanism of how the SH2/SH3/SH2 region positively influences GAP activity remains to be determined, sequences within this region may specifically interact with the catalytic domain to stimulate activity. Alternatively, it is possible that the SH2/SH3/SH2 region contributes to the overall conformational stability of the protein, and thus large deletions in this region may cause nonspecific conformational changes that alter catalytic function. Further study will be required to distinguish between these possibilities. Nevertheless, in addition to the many other roles that these domains have, our findings demonstrate that the SH2 and SH3 domains are essential for p120 GAP to display full catalytic activity toward Ras. These results are of particular interest in light of the observation that mutations in the gene encoding p120 GAP have been detected in human basal cell carcinomas (22). Significantly, in all three cases examined, point mutations resulting in amino acid changes were detected in the C-terminal SH2 domain but not in the catalytic domain of p120 GAP. Our in vitro findings raise the possibility that these p120 GAP mutants have reduced catalytic activity, which would lead to activation of Ras and hence contribute to oncogenesis. We cannot exclude, however, that the SH2 mutations influence p120 GAP intracellular location or interaction with other proteins in vivo and in this way contribute to oncogenesis. Furthermore, it should be noted that earlier studies (3) suggested that the isolated catalytic domain of p120 GAP was at least as efficient as the full-length protein in suppressing cell transformation by Src or Ras. While the basis for this apparent discrepancy with our findings is not clear, these contrasting results suggest that the regulation of p120 GAP in intact cells may be complex. It will be of significant interest to characterize the in vitro and in vivo properties of the C-terminal SH2 domain mutants associated with basal cell carcinomas (22). Based on our biochemical data, mutations in the N-terminal SH2 or SH3 domains might also contribute to human cancers, providing a rationale for screening tumors for mutations in the entire SH2/SH3/SH2 region of p120 GAP.