Tyrosine phosphorylation of Ras GTPase-activating protein stabilizes its association with p62 at membranes of v-Src transformed cells.

Ras GTPase-activating protein (GAP) regulates the activity of Ras proteins, which have key roles in signal transduction pathways downstream of oncogenic and receptor tyrosine kinases. Previous studies indicated that Tyr-457 of bovine GAP (Tyr-460 of human GAP) is the major site of phosphorylation by viral Src (v-Src) kinase and epidermal growth factor receptor. The finding that Tyr-457 in GAP is located immediately adjacent to Src homology 2 (SH2) and 3 (SH3) domains led us to investigate the possibility that this specific phosphorylation regulates protein-protein interactions involving GAP. For this purpose, we constructed a full-length GAP mutant containing a substitution of Phe-457 in place of Tyr-457. Both wild-type GAP and mutant GAPF457 were tagged with the KT3 epitope at the carboxyl terminus and were expressed in v-Src transformed rat fibroblasts. In vivo phosphorylation analyses established that GAPF457 was weakly phosphorylated on tyrosine and, as expected, lacked the phosphopeptide containing Tyr-457. Analysis of GAP-associated proteins in anti-KT3 immunoprecipitates showed that GAP stably associated with two major phosphoproteins, p62 and p190, which have been previously described. Significantly, association of p62 with GAPF457 was reduced approximately 3-fold compared with wild-type GAP. Subcellular fractionation experiments further demonstrated that Tyr-457 phosphorylation of GAP stabilized its association with p62 at cell membranes. Based on these findings, we propose that one role of tyrosine phosphorylation in GAP is to enhance its association with p62 at membranes, which in turn may contribute to regulation of Ras signal transduction pathways.

Tyrosine Phosphorylation of Ras GTPase-activating Protein Stabilizes Its Association with p62 at Membranes of v-Src Transformed Cells* (Received for publication, July 7, 1993, and  in GAP is located immediately adjacent to Src homology 2 (SH2) and 3 (SH3) domains led us to investigate the possibility that this specific phosphorylation regulates protein-protein interactions involving GAP. For this purpose, we constructed a full-length GAP mutant containing a substitution of Phe-457 in place of Tyr-457. Both wild-type GAP and mutant GAPF467 were tagged with the KT3 epitope at the carboxyl terminus and were expressed in v-Src transformed rat fibroblasts. In vivo phosphorylation analyses established that GAPF467 was weakly phosphorylated on tyrosine and, as expected, lacked the phosphopeptide containing Tyr-457. Analysis of GAP-associated proteins in anti-KT3 immunoprecipitates showed that GAP stably associated with two major phosphoproteins, p62 and p190, which have been previously described. Significantly, association of p62 with GAPF467 was reduced approximately %fold compared with wild-type GAP. Subcellular fractionation experiments M h e r demonstrated that Tyr-457 phosphorylation of GAP stabilized its association with p62 at cell membranes. Based on these findings, we propose that one role of tyrosine phosphorylation in GAP is to enhance its association with p62 at membranes, which in turn may contribute to regulation of Ras signal transduction pathways.
Ras protein has a central role in regulation of normal cell proliferation and in oncogenesis (1). Ras functions like a GTPbinding regulatory protein (G-protein) with a GTP/GDP cycle that is controlled by other proteins (1-4). Normal cellular Ras is biologically and biochemically active only when bound to GTP, and oncogenic activation of Ras is induced by mutations that either inhibit GTP hydrolysis or facilitate GTPIGDP exchange (2-5). The current hypothesis regarding signalling pathways involving Ras is that the Ras-GTP complex binds to effector molecules and transmits signals before GTP hydrolysis returns Ras to the inactive GDP complex ( 2 4 , 6). Much evi-dence indicates that Ras acts downstream of tyrosine kinases, including viral Src (v-Src),l in mitogenic signalling pathways and that tyrosine kinases can activate Ras function (6-10). However, the mechanism by which this Ras activation by upstream tyrosine kinases occurs is not completely understood.
Ras GTPase-activating protein (GAP) was discovered by its ability to stimulate the intrinsic GTPase activity of Ras by over 100-fold (11). GAP interacts with cellular Ras only in the active GTP-bound form and inactivates Ras by stimulating its GT-Pase activity (2,3,(11)(12)(13). This negative regulatory function of GAP is consistent with the finding that overexpression of human GAP blocks oncogenic transformation by cellular Ras (14).
In addition to a regulatory function, GAP may be required for Ras effector function, possibly as one of the downstream target molecules of Ras (2,4,15). Recent studies also point to the important regulatory role of guanine-nucleotide releasing proteins, which convert the inactive Ras-GDP complex to the active GTP complex by stimulating nucleotide exchange (16-21). Mitogenic signal transduction through Ras, therefore, involves stimulation of nucleotide exchange factors in addition to GAP function.
Ras GTPase-enhancing activity of GAP resides in the carboxyl-terminal region (22), while the amino-terminal portion contains Src homology 2 (SH2) and 3 (SH3) regions (13). SH2 regions have been shown to bind phosphorylated tyrosine residues in proteins and are implicated in the interactions of tyrosine kinases with their targets (23-25). GAP is phosphorylated by a variety of oncogenic tyrosine kinases, including v-Src, as well as by the ligand-stimulated receptor tyrosine kinases (26,27). GAP also forms complexes with the platelet-derived growth factor receptor and Src kinase, interactions which can be mimicked in vitro by using either baculoviral expressed GAP or bacterially expressed GAP SH2 domains (24, 28-31). These findings suggest that GAP might serve as a link between tyrosine kinases and Ras in mitogenic signalling pathways.
GAP forms complexes with two additional phosphoproteins, p62 and p190, in cells expressing activated tyrosine kinases, including v-Src (27, 32). The isolated GAP SH2 domains expressed in bacteria can form stable complexes in vitro with p62 (25), suggesting that GAP complex formation with p62 is a specific SH2-mediated interaction. Activation of tyrosine kinases may induce the formation of signalling complexes involving GAP, p62, and p190, which could in turn regulate signal transduction through Ras. Recently, we and others have shown that Tyr-457 of bovine GAP (corresponding to Tyr-460 of human GAP) is the major tyrosine residue phosphorylated by both activated Src kinase and the epidermal growth factor receptor kinase (33,34). Because the Tyr-457 phosphorylation site is The abbreviations used are: v-Src, viral Src; GAP, GTPase-activating protein.

25728
located immediately COOH-terminal to one of two SH2 regions in GAP, this specific phosphorylation might regulate proteinprotein interactions, particularly GAP association with p62. Here we report that tyrosine phosphorylation of GAP stabilizes its association with p62 at membranes of cells transformed by v-Src. These findings suggest that one function of GAP phosphorylation is to regulate its interaction with p62, which in turn may contribute to regulation of Ras mitogenic signalling pathways.

MATERIALS AND METHODS
Cell Culture-Rat 3Y1 fibroblasts transformed by Rous sarcoma virus (SR-3Y1) (35) were maintained in Dulbecco's modified Eagle's medium containing iron-supplemented 5% bovine calf serum (Hyclone) as described earlier (31,34). Dexamethasone (Sigma) treatment of cells was performed for 15-20 h and was used at a concentration of 300 m.
Construction and Expression of KT3-tagged GAP-For introducing the 8-amino acid KT3 epitope (TPPPEPET) (36) to the COOH terminus of bovine GAP, a double-stranded 55-base pair oligonucleotide was synthesized that encodes the extreme COOH-terminal 8 amino acids of GAP plus the KT3 tag, flanked by an AccI site at its 5'-end and a TAG stop codon plus EcoRI site at its 3'-end. The oligonucleotide was phosphorylated and cloned into the EcoRI and XbaI sites of pUC18 together with an XbaI-AccI GAP fragment (nucleotides 2947-3105 of the bovine GAP coding region) (37).
For efficient expression of bovine GAP cDNA in rat 3Y1 cells, a double-stranded 83-base pair oligonucleotide was synthesized consisting of the 5'-GAP coding sequence in addition to a modified 5"untranslated sequence conforming to Kozak's rules (38). flanked by a Sa21 site at its 5'-end and a Not1 site at its 3'-end. The phosphorylated oligonucleotide was cloned into the BamHI and SalI sites of pUC18 together with a NotI-BamHI GAP fragment (nucleotides 64-2098 of bovine GAP). For regenerating the full-length GAP construct, the SalI-BamHI fragment and XbaI-EcoRI fragment cloned into pUC18 vectors described above were isolated and cloned into the SalI and EcoRI site of pUC18 together with a BamHI-XbaI GAP fragment (nucleotides 2098-2947 of bovine GAP). Modified sequences at both the 5'-end and 3'-end of the full-length GAP construct were confirmed by the dideoxy DNA sequencing method (39).
To generate full-length GAPF457, a 601-base pair PstI fragment of bovine GAP containing the codon for Tyr-457 was isolated and subcloned into pUC18. Site-specific mutagenesis was performed by the sequential polymerase chain reaction method (40) with pfu polymerase (Strategene), using two oligonucleotide primers incorporating TlT (Phe) instead of TAT (Tyr) plus universal reverse and forward primers. Amplified polymerase chain reaction products were further digested with PstI and BglII and the isolated GAP fragment (nucleotides 1101-1548 of bovine GAP) was subcloned into the full-length GAP in pUC18. DNA sequencing in the region of the polymerase chain reaction mutagenesis was performed to exclude the possibility of errors by the polymerase. Structures of the GAP recombinants described above are shown in Fig. 1.
SalI-EcoRI fragments containing the full-length GAP or GAPF457 coding sequence from the above constructs were blunt-ended and cloned into SalI-restricted and blunt-ended pMAMneo (Clonetech) expression vector. Wild-type GAP and GAPF457 pMAMneo constructs, and p M A " neo vector alone control, were transfected using the calcium phosphate coprecipitation method (41) into SR-BY1 cells. G418-resistant colonies were isolated as described earlier (42) and cell lines expressing KT3tagged GAP in response to dexamethasone were further expanded.
Lysates were clarified by centrifugation in a microcentrifuge for 15 min at 4 "C, and then proteins were immunoprecipitated by incubating with antibodies for 1 h at 4 "C as previously described (31, 34). Protein A-Sepharose (Pharmacia) was then added for 30 min and immunoprecipitates were washed three times by pelleting in RIPA buffer without protease inhibitors. Immunoprecipitates were boiled in SDS sample buffer, loaded on 7.5% polyacrylamide-SDS gels, and separated by electrophoresis.
To disrupt protein-protein complexes, cells were lysed in SDS buffer The 8-amino acid KT3 epitope tag, which is fused to each COOH-terminal end, is shown as vertical stripes. Proteins were expressed from the pMAMneo vector under the control of the mouse mammary tumor virus long terminal repeat (see "Materials and Methods" for details of recombinant DNA constructs).
Proteins resolved by SDS-PAGE were electrophoretically transferred to nitrocellulose and probed with the indicated antibodies as previously described (31,34). Monoclonal antibody probes were detected using 10 pCi of 1z51-labeled sheep anti-mouse IgG (ICN) followed by exposure to Kodak XAR-5 film with an intensifjmg screen at -80 "C. For immunoblots probed with polyclonal antibodies, the bound antibodies were detected using 10 pCi of lz5I-1abeled affinity purified protein-A (Amersham). Quantitation of 1251-labeled IgG or protein-A binding to immunoblots was performed with a Molecular Dynamics PhosphorImager.
Subcellular Fractionation-Cells were scraped into cold phosphatebuffered saline containing 1 m sodium orthovanadate and then collected by centrifugation. 8 g of cell pellets (wet weight) were resuspended with 3 ml of hypotonic buffer (10 nm Tris-C1, pH 7.5, 1 m MgCI,, 1 m phenylmethanesulfonyl fluoride, 1 m sodium orthovanadate, 1 w leupeptin, 1 1.1~ antipain, and 0.1 w aprotinin) (43); 10 min later, the cells were transferred to a Dounce homogenizer and disrupted by 40 strokes. The homogenate was centrifuged at 800 x g for 10 min to remove nuclei and then the postnuclear supernatant was centrifuged at 100,000 x g for 40 min. The supernatant fraction (S100) was adjusted to 1% Nonidet P-40. The pellet (P100) was gently rinsed with 3 ml of phosphate-buffered saline and then resuspended in 3 ml of hypotonic buffer containing 1% Nonidet P-40; 10 min later, insoluble materials were removed by centrifugation in a microcentrifuge for 5 min. The SlOO and PlOO fractions were immunoprecipitated with anti-KT3 antibodies and washed two times with cold RIPA buffer containing 1 m sodium orthovanadate as described above.
In Vivo Labeling and Phosphopeptide Mapping Analysis-For metabolic labeling, SR-3Y1 cells seeded in 10-cm dishes were grown to confluence and treated with dexamethasone prior to labeling. Cellular proteins in these cells were labeled with 5 mCi of [32Plorthophosphate (ICN) by incubation for 3 h at 37 "C in 2 ml of phosphate-free Dulbecco's modified Eagle's medium (Life Technologies Inc.) supplemented with 0.5% fetal calf serum plus 50 w sodium orthovanadate.
Labeled proteins separated by SDS-PAGE were transferred to nitrocellulose membranes and then localized by autoradiography. Immobilized proteins were directly digested with 20 pg of trypsin (Worthington) by incubating membrane pieces in 50 m NH,HC03 (pH 7.3-7.6) at 37 "C for 4 h and tryptic peptides were then oxidized with performic acid. Tryptic peptides were analyzed by electrophoresis in pH 1.9 buffer (first dimension) and chromatography (second dimension) on TLC plates. All of the procedures involving peptide mapping analysis were performed according to previously described protocols (44.45).

RESULTS
More Than 10% of Total Cellular GAP Is Qrosine Phosphorylated in v-Src lkansformed Rat Fibroblasts-Previous stud-ies established that GAP is tyrosine phosphorylated in v-Src transformed cells (27, 301, although the stoichiometry of this modification has not been determined. To estimate the fraction of tyrosine-phosphorylated GAP in v-Src transformed SR-3Y1 fibroblasts, cell lysates were prepared using RIPA buffer for a n t i -r n immunoprecipitation followed by anti-GAP immunoblot analysis. This analysis revealed that only 1% of total cellular GAP was recovered by anti-FTyr antibodies (data not shown). Earlier studies demonstrated that tyrosine-phosphorylated GAP physically associates with two major phosphoproteins, p62 and p190, in v-Src transformed cells (27,32,43). If phosphorylation is important for these protein-protein interactions, phosphotyrosine in GAP could be inaccessible to anti-PTyr antibodies. Therefore, cells were lysed under denaturing conditions in boiling SDS buffer that disrupted GAP association with p190 and p62. Denatured lysates were subsequently diluted 10-fold in RIPA buffer prior to immunoprecipitation with anti" antibodies. The relative amount of GAP recovered by anti-PTyr antibodies was compared with the amounts of total cellular GAP from the same lysate by immunoblot analysis using anti-GAP antibodies as probe (Fig. 2). For maximum recovery of tyrosine-phosphorylated GAP, three consecutive immunoprecipitations of the same lysate were combined and analyzed (Fig. 2, lane 4 ) . Results indicated that about 12% of total cellular GAP was recovered by anti-FTyr antibodies (Fig. 2,  compare lanes 2, 3, and 4 ) . The lysate supernatant after the third immunoprecipitation was further immunoprecipitated with anti" antibodies ( Fig. 2. lane 5) to confirm that most tyrosine-phosphorylated GAP was recovered in the previous immunoprecipitations. As a control, three consecutive immunoprecipitations using anti-PTyr antibodies preincubated with phenylphosphate as competitor were combined and analyzed to demonstrate that tyrosine-phosphorylated GAP was specifically immunoprecipitated by anti" antibodies ( Fig. 2. compare lanes 4 and 6). In several independent experiments, we observed that between 10 and 20% of endogenous cellular GAP in SR-3Y1 cells was recovered by a n t i -w antibodies, indicating that at least 10% of cellular GAP is tyrosine phosphorylated in v-Src transformed rat fibroblasts.  (lanes 4 and 6, respectively). The tated once more with a n t i -m antibodies (lane 5). Proteins recovered supernatant after the third immunoprecipitation was immunoprecipiby a n t i -m antibodies were combined and analyzed by immunoblot using anti-GAFY638) antibodies as probe followed by affinity-purified '261-labeled protein -A (lanes 4-6). For comparison. 5, 10, and 15% aliqouts of the undiluted, whole lysate prior to immunoprecipitation were directly analyzed on the same blot (lanes 1-3).

GAPF'b' Expressed in v-Src lhnsfonned Rat Fibroblasts Lacks a Phosphopeptide That Contains 2)r-457-%vious
studies suggested that Tyr-457 in bovine GAP is the major site phosphorylated in vitm and in vivo by the v-Src and epidermal growth factor receptor kinases (33,34). To confirm that Tyr-457 in GAP is the major phosphorylation site in v-Src transformed cells, full-length GAP containing Phe-457 instead of Tyr457 was constructed as described under 'Materials and Methods." At the COOH terminus of each GAP construct, an 8-amino acid KT3 epitope corresponding to the COOH terminus of SV40 large T-antigen (36) was appended to facilitate immunoprecipitation of exogenously expressed GAP by anti-KT3 monoclonal antibody (Fig. 1). These GAP constructa were introduced into v-Src transformed cells and stable cell lines were established that expressed from 1.5-to 2-fold more exogenous GAP than endogenous GAP in response to dexamethasone. No changes in cell morphology were apparent in these v-Src transformed cells irrespective of GAP induction (data not shown). Lysates from each stably transfected cell line incubated with or without dexamethasone were prepared for immunoprecipitation with anti-KT3 antibodies followed by immunoblot analysis using anti-GAP antibodies as probe. KT3-tagged GAP or GAPF467 was inducibly expressed by dexamethasone, although a low level of KT3-tagged GAP was expressed in the absence of dexamethasone (Fig. 3A). No endogenous GAP was immunoprecipitated from the control cell lysate by anti-KT3 antibodies,

SR-3Y1 cells stably transfected with pMAVneo
vector, or wild-type GAP or GAPp467 construcL9 derived from this vector. were untreated or treated with 300 rn dexamethasone for 15 h prior to preparing cell lysates and labeling with ["PIorthophosphate.A, unlabeled cell lysates were immunoprecipitated with anti-KT3 antibodies. and then analyzed by immunoblot using anti-GAP(638) antibodies a s probe followed by affinity-purified '2nI-labeled protein A. B , cells were metabolically labeled with ["'Plorthophosphate for 3 h. Labeled KT3-tagged GAP proteins were isolated by immunoprecipitation with anti-KT3 antibodies, separated by SDS-polyacrylamide gel electrophoresis and then transferred to a nitrocellulose membrane. GAP proteins were localized by autoradiography, and then tryptic phosphopeptides were prepared a~ described under 'Materials and Methods." 'IYyptic phosphopeptides were analyzed on TLC plates by electrophoresis a t pH 1.9 in the horizontal dimension with the anode on the le/?. and ascending chromatography in the vertical dimension. The origin in marked with an arrow. and the phosphopeptides derived from GAP are labeled A and B .
demonstrating that anti-KT3 antibody did not cross-react with endogenous GAP (Fig. 3A, lanes I and 2).
To determine whether GAPF457 lacks the major phosphorylation site in v-Src transformed cells, each stably transfected cell line was metabolically labeled with [32P]orthophosphate in the presence of dexamethasone. Immunoprecipitates prepared from these cell lysates with anti-KT3 antibodies contained phosphorylated GAP or weakly phosphorylated GAPF467 (data not shown; see below). 32P-Labeled GAP or GAPF457 was isolated, digested with trypsin, and analyzed by two-dimensional phosphopeptide mapping. As expected from previous studies (34), wild-type GAP contained two tryptic phosphopeptides, A and B, but GAPF4"' contained only tryptic phosphopeptide B (Fig. 3B). Because our earlier studies established that phosphopeptide A contains "457 as the only phosphorylation site (34), these results confirm that "'37-457 is the major in vivo phosphorylation site in v-Src transformed cells. Other phosphopeptides that are variably detected appear to arise from phosphoproteins nonspecifically immunoprecipitated by anti-KT3 antibodies because these phosphopeptides are not observed using in vivo labeled endogenous GAP immunoprecipitated by anti-GAP antibodies (data not shown).
Binding of p62 to Is Reduced Compared with Wildtype GAP-Tyr-457 in bovine GAP is located immediately COOH-terminal to the second SH2 region (Fig. 11, suggesting that phosphorylation of this site might regulate protein-protein interactions, particularly GAP association with p62. To test this hypothesis, v-Src transformed rat 3Y1 cells stably transfected with vector, GAP, or GAPF457 constructs were treated with dexamethasone to induce KT3-tagged GAP expression prior to preparing cell lysates. Immunoprecipitates prepared with anti-KT3 antibodies were analyzed directly by immunoblot using either a n t i -m or anti-GAP antibodies as probes  scribed in the legend to Fig. 4. Quantitation of lmI-labeled I@ or a Each of the three independent experiments was performed as deprotein A bound to immunoblota was done with a PhosphorImager. The amount of GAP-associated p62. quantitated with antiphosphotyrosine antibodies, was normalized to the amount of GAP, quantitated with anti-GAP antibodies, prior to calculating the relative percentage of pf~2 bound to GAP. The amounts of wild-type GAP-associated p62 were assumed to be 1001. ( Fig. 4). Similar amounts of KT3-tagged GAP were present in lysates from cells expressing GAP or GAPF4K7, but none was detectable in control cell lysate (Fig. 4, bottom panel ). As expected from the in vivo phosphopeptide mapping experiments described above, tyrosine phosphorylation of was significantly reduced because it lacked the major phosphorylation site (Fig. 4, top panel ). Nevertheless, the overall levels of p190 coimmunoprecipitated with GAP or GAPp4K7 were not substantially changed, indicating that association of GAP with p190 did not involve " 4 5 7 phosphorylation of GAP (Fig. 4, top panel 1. In some experiments, however, we variably observed that GAPNs7 associated with more p1W than did wild-type GAP (data not shown). In contrast, we reproducibly found that the amounts of p62 coimmunoprecipitated with GAP correlated with the degree of GAP tyrosine phosphorylation (Fig. 4), suggesting that Tyr-457 phosphorylation of GAP is important for ita association with p62 protein. Quantitation shown in Table I using a PhosphorImager indicated that the relative binding of p62 to GAPF457 decreased approximately &fold compared with tyrosine-phosphorylated wild-type GAP.
Binding of p62 to GApF"7 Is Greatly Reduced at Cell Membranes-Previous studies established that a substantial fraction of tyrosine-phosphorylated GAP is localized at cell membranes in response to growth factor stimulation or in v-Src transformed cells (26,43). To determine whether GAP-p62 mmplexes are localized to membranes as a function of "yT-457 phosphorylation in GAP, cell lysates were separated into membrane and cytoplasmic fractions. Immunoprecipitates prepared with anti-KT3 antibodies from each fraction were analyzed directly by immunoblot using either anti" or anti-GAP antibodies as probes (Fig. 5A). The amounts of wild-type GAP (-25%) associated with cell membranes were reproducibly more than GAPF467 (-15%) (Fig. 5, A, bottom panel, and B ), indicating that localization of GAPp4K7 to membranes is partially defective. Tyrosine-phosphorylated wild-type GAP is distributed more in the particulate fraction than the soluble fraction, while GAPF'457 phosphorylated at the minor site of tyrosine phosphorylation is present predominantly in the soluble fraction, indicating that w-457-phosphorylated GAP is enriched at cell membranes (Fig. 5A, top panel ). In repeated experiments, we observed that some tyrosine-phosphorylated proteins comigrating with GAP were nonspecifically immunoprecipitated by anti-KT3 antibodies (data not shown). Therefore, although the anti-PTyr immunoblot reflects the distribution of tyrosine-phosphorylated GAP, this analysis does not show the absolute amounts of tyrosine-phosphorylated GAP.
Wild-type GAP-associated p62 was distributed slightly more in the particulate fraction than soluble fraction (Fig. 5, A, top  panel, and B ) , consistent with a previous report (43). Subcellular distribution of the wild-type GAP-associated p62 con" sponded to the distribution of tyrosine-phosphorylated GAP rather than total cellular GAP ( Fig. !  5 A , compare lanes 3 and 4 ), suggesting that Tyr-457 phosphorylation of GAP enhances GAP s association with p62. In striking contrast, we reproducibly verely reduced at cell membranes and enriched in the soluble fraction (Fig. 5,A, top panel, and B 1. These results indicate that Tyr-457 phosphorylation of GAP stabilizes its interaction with type GAP or GAPF467 was predominantly in the cytoplasmic fraction (Fig. 5A, top panel, lanes 4 and 6). Subcellular distribution of p190 corresponds to the distribution of total cellular GAP ( Fig. 5 A ) , suggesting that p190 associates mainly with unphosphorylated GAP. The amounts of GAPF457-associated p190 in the cytoplasmic fraction were detectably more than wild-type GAP-associated p190 (Fig. 5A. compare lanes 4 and   6). These results suggest that the stoichiometry of plW-GAP Ir found that the amounts of GAP'4s7-associated p62 were se-P ? S P S 6% --* -. " -m p62 at cell membranes. Unlike p62, p190 associated with wild-GAP " a p e 2 complexes may be influenced by reduced p62-GAP association.

anti-PTyr DISCUSSION
Previous studies showed that GAP is one of the Droteins that is tyrosine phosphorylated in v-SIT transformed cells (27,30). In addition, our in vitro reconstitution experiments using a baculovirus expression system demonstrated that GAP is a direct substrate for the v-Src tyrosine kinase (31). Because GAP is a key regulator of Ras, which is implicated in mitogenic signalling pathways downstream of v-Src (8). these findings suggest that GAP might be a physiologically relevant substrate for the SIT tyrosine kinase. In this study, we measured the approximate stoichiometry of tyrosine-phosphorylated GAP in v-Src transformed rat fibroblasts. Using a denatured lysate procedure with a n t i -m antibodies and appropriate controls, we determined that more than 10% of t o t a l cellular GAP is tyrosine phosphorylated in these v-Src transformed cells. However, this is only a minimum estimate because phosphotyrosine could be unstable during cell lysis and the anti" antibodies might not react with all of the phosphorylated GAP. Other studies using platelet-derived growth factor-stimulated NIH 3T3 cells indicated that the stoichiometry of tyrosine-phosphorylated GAP was approximately 30% in those cells (47). These results suggest that a substantial fraction of cellular GAP is a direct target for activated tyrosine kinases, consistent with a significant physiologic role of tyrosine phosphorylation in GAP.
Our earlier studies suggested that 2 tyrosine residues in GAP are phosphorylated in v-SIT transformed cells, and that one of these is "457 (of bovine GAP) which is directly phosphorylated by v-SIT (34). The other unidentified tyrosine phosphorylation site is minor and appears to be phosphorylated by another kinase present in v-Src transformed cells (34). Our phosphopeptide mapping experiments of GAPF4" reported here confirm that Tyr-457 of bovine GAP is the major in vivo phosphorylation site in v-SIT transformed cells. Significantly, the equivalent of Tyr-457 in human GAP is specifically phosphorylated in cells stimulated by epidermal growth factor (33), providing further support for the suggestion that this modification is physiologically relevant.
GAP has two SH2 domains and one SH3 domain, in the order SH2-SH3-SH2. Previous in vitro studies using the NH2-terminal GAP SH2 domain (SH2-N) expressed in bacteria indicated that GAP undergoes SH2-mediated interactions with the epidermal growth factor receptor, platelet-derived growth factor receptor, and p62 (24.25). Extensive mutagenesis in the GAP associated p62 (bottom panel ), were quantitated by using a Phwphorparticulate or soluble fractions are indicated by percentage. Solid har Imager. For comparison, the relative amounte of proteins in either hatched bar indicates GAP or associated $2 in the soluble fraction.
represents GAP or associated p62 in the particulate fraction and Each data set represents the mean of three separate experimentn S.D.
SH2-N domain demonstrated that highly conserved basic amino acids are important for p62 binding to GAP in vitro (48).
These results suggest that the SH2-N domain in GAP is a binding site for p62. In comparison to the GAP SH2-N domain, the GAP SH2-C domain binds weakly to tyrosine-phosphorylated proteins in vitro (24,25). While the role of the GAP SH2-C domain is not clear, it has been implicated in cooperative binding of both GAP SH2 domains to tyrosine-phosphorylated proteins in vivo (24,25,48). The finding that Tyr-457 in bovine GAP is located immediately COOH-terminal to the SH2 domains led us to postulate that this specific phosphorylation might regulate GAP association with p62. Our results presented here suggest that the binding of p62 to the GAPF457 mutant, which lacks the major phosphorylation site, was decreased approximately %fold compared with wild-type GAP in v-Src transformed cells. Because p62 was detected by anti" antibodies, however, these data do not distinguish between decreased p62 bound to GAP or decreased phosphorylation of GAP-associated p62. One role of tyrosine phosphorylation in GAP, therefore, might be to increase its binding affinity for tyrosine-phosphorylated p62 by modulating the SH2-SH3-SH2 region conformation in GAP. Other studies have suggested that, in murine fibroblasts overexpressing c-Src, GAP is not tyrosine phosphorylated and yet it associates with p62 (49). These results are consistent with our finding that blocking GAP phosphorylation does not completely abolish its association with p62, indicating that some other factor(s) also contribu t e (~) to their interaction.
Previous studies using subcellular fractionation procedures established that GAP is almost entirely cytosolic in normal cells, and activated tyrosine kinases induce a redistribution in GAP subcellular localization to membranes (26,431. However, it should be noted that the specific cellular membranes with which tyrosine-phosphorylated GAP is associated remain to be determined. Consistent with these reports, we observed that up to 25% of GAP i s localized a t membranes of cells transformed by v-Src. Once a t cell membranes, GAP may be phosphorylated on Tyr-457 by the v-Src kinase and then become stably associated with tyrosine-phosphorylated p62. Indeed, our results indicate that association of tyrosine-phosphorylated wild-type GAP with p62 is significantly enhanced at cell membranes compared with the cytoplasm. In the case of GAPF457, blocking phosphorylation of Tyr-457 partially reduced the amount of GAP a t membranes and, more strikingly, shifted the majority of GAP-associated p62 to the cytoplasm. Our studies also suggest that the stoichiometry of GAP-pl90 complexes is influenced to some extent by Tyr-457 phosphorylation in GAP. One explanation for why GAPF457 bound to more p190 compared with wildtype GAP is that more GAPF457 could be available for binding to p190 due to its reduced association with p62. Because the binding properties of p190 have not been fully characterized, however, it is difficult to determine whether p190 and p62 might be competing for overlapping binding sites on GAP. In summary, our results provide the first demonstration of a biochemical consequence of GAP tyrosine phosphorylation in vivo. Tyrosine phosphorylation of GAP correlates well with association of p62 but not with association of p190. While p62 association with GAP involves GAP SH2 domains and phosphotyrosines in p62, Tyr-457 phosphorylation in GAP stabilizes its association with p62, especially a t cell membranes. Our findings are consistent with a model in which phosphorylation of GAP by activated tyrosine kinases stimulates Ras signalling pathways through increased GAP association with p62 a t membranes. Since Ras is localized to cell membranes, it is possible that p62 regulates GAP GTPase-enhancing activity towards Ras or contributes to an effector function of GAP in response to Ras activation. Recent evidence that the NH,-terminal portion of GAP containing the SH2 domains induces gene expression in a Ras-dependent manner (50) may support the suggestion of an effector role for the p62-GAP complex in mitogenic signal transduction. A cDNA encoding p62 has been cloned and the predicted protein shares extensive sequence similarity with an hnRNP protein (51). Because our results suggest that the association of p62 with GAP is enhanced by GAP tyrosine phosphorylation at cell membranes, further investigation of p62 function may provide new insights into Ras-mediated signal transduction. 14.