The Novel Role of the C-terminal Region of SHP-2

SHP-2, a nontransmembrane-type protein-tyrosine phosphatase that contains two Src homology 2 (SH2) domains, is thought to participate in growth factor signal transduction pathways via SH2 domain interactions. To determine the role of each region of SHP-2 in platelet-derived growth factor signaling assayed by Elk-1 activation, we generated six deletion mutants of SHP-2. The large SH2 domain deletion SHP-2 mutant composed of amino acids 198–593 (SHP-2-(198–593)), but not the smaller SHP-2-(399–593), showed significantly higher SHP-2 phosphatase activity in vitro. In contrast, SHP-2-(198–593) mutant inhibited wild type SHP-2 phosphatase activity, whereas SHP-2-(399–593) mutant increased activity. To understand these functional changes, we focused on the docking protein Gab1 that assembles signaling complexes. Pull-down experiments with Gab1 suggested that the C-terminal region of SHP-2 as well as the SH2 domains (N-terminal region) associated with Gab1, but the SHP-2-(198–593) mutant did not associate with Gab1. SHP-2-(1–202) or SHP-2-(198–593) inhibited platelet-derived growth factorinduced Elk-1 activation, but SHP-2-(399–593) increased Elk-1 activation. Co-expression of SHP-2-(1–202) with SHP-2-(399–593) inhibited SHP-2-(399–593)/Gab1 interaction, and the SHP-2-(399–593) mutant induced SHP-2 phosphatase and Elk-1 activation, supporting the autoinhibitory effect of SH2 domains on the C-terminal region of SHP-2. These data suggest that both SHP-2/Gab1 interaction in the C-terminal region of SHP-2 and increased SHP-2 phosphatase activity are important for Elk-1 activation. Furthermore, we identified a novel sequence for SHP-2/Gab1 interactions in the C-terminal region of SHP-2.

SHP-2 is a widely expressed protein-tyrosine phosphatase that has two tandem SH2 1 domain repeats at its N terminus (1). SHP-2 binds directly to growth factor receptors, including the PDGF and EGF receptors, in response to receptor stimula-tion with the corresponding ligand and undergoes tyrosine phosphorylation (2,3). Accumulating evidence implicates SHP-2 as a positive regulator of ERK activity downstream from receptor-tyrosine kinases (4). Fibroblasts derived from SHP-2 exon 3Ϫ/Ϫ mice exhibit decreased ERK activity in response to EGF and PDGF. Following growth factor stimulation, SHP-2 is recruited through its SH2 domain directly to the EGF or PDGF receptor (5,6). SHP-2 function is regulated by both intramolecular associations and binding of regulatory proteins (Fig.  1A). Schematically, SHP-2 consists of a phosphatase domain in the C terminus (PTP2 in Fig. 1A) and two tandem SH2 domains in the N terminus. It has been proposed that the SH2 domains maintain SHP-2 in an inactive conformation under basal conditions. Specifically, previous investigators found that deletion of the SH2 domains of SHP-2 protein enhanced phosphatase activity, and binding of tyrosine-phosphorylated peptides to the SH2 domains (presumably removing their inhibitory interaction with PTP2) increased SHP-2 catalytic activity (7,8). The SHP-2 crystal structure provides additional support for this model (9), since it predicts that the N-terminal SH2 domain binds the phosphatase domain and directly blocks its active site.
In addition to SH2 domains, it is possible that the C-terminal portion of SHP-2 associates with other proteins via putative phosphotyrosine sites that may interact with SH2 domain (1). Interestingly, SHP-2 is activated by phospholipids that may be mediated by lipid interaction with the phosphatase domain, because a deletion mutant of SH2 domains still can respond to phospholipid stimulation (7). Recently, Xu et al. reported that SHP-2 co-localized with stress fibers at low cell densities and directly associates with F-actin through the phosphatase domain (10).
Based on previous data, there are three possible models by which SHP-2 may regulate ERK1/2 or Elk-1 activity. 1) Both SHP-2-Gab1 complex formation and SHP-2 phosphatase activity are critical for ERK1/2 and Elk-1 activation. 2) The Nterminal SH2 domains of SHP-2 associate with Gab1, which increases SHP-2 phosphatase activity and activates ERK1/2 and Elk-1 activity. 3) Association of Gab1 with SHP-2 is the only mechanism necessary to regulate ERK1/2 and Elk-1 ac-tivity, and SHP-2 phosphatase activity is not directly involved in this process. Model 1 is supported by data indicating that both a catalytic inactive SHP-2 mutant (SHP-2C/S) and disruption of Gab1/SHP-2 binding inhibit ERK1/2 activation, but the relationship between SHP-2 phosphatase activity and SHP-2/ Gab1 interaction remains unclear. Model 2 is supported by the structure of SHP-2 (9). However, since SH2 domain deletion mutants of SHP-2, which act as a constitutively active form of SHP-2, inhibit ERK1/2 activity (4,7,8), the actual involvement of SHP-2 phosphatase activity is not clear. Model 3 is supported by the data that the mutation of SHP-2 binding sites in Gab1 abolished growth factor-induced ERK1/2 activation as described above. In addition, since a catalytically inactive form of SHP-2 (SHP-2C/S) is able to form a complex with Gab1 constitutively (17), a possible mechanism for SHP-2C/S inhibition of ERK1/2 activity may be "trapping" Gab1 from SHP-2 association.
In the current study, we made several deletion mutants of SHP-2 to clarify the relationship between SHP-2 phosphatase activity and SHP-2/Gab1 interaction. We found that the SHP-2-(399 -593) mutant associated with Gab1. Interestingly, the SHP-2-(399 -593) mutant increased wild type SHP-2 phosphatase activity and Elk-1 activity. In contrast, other SHP-2 mutants, which had either no association with Gab1 or no phosphatase activity, significantly inhibited PDGF-induced Elk-1 activity. These results suggest that both Gab1/SHP-2 interaction and SHP-2 phosphatase activity are important for SHP-2 signaling events such as Elk-1 activation. Structure function analysis indicates a novel role for the C-terminal region of SHP-2 in regulating Elk-1 activation by both mediating interaction with Gab1 and regulating SHP-2 phosphatase activity.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP was purchased from Amersham Biosciences. Raytide and pp60 c-Src tyrosine kinase were from Calbiochem. Anti-hemagglutinin (HA), polyclonal SHP-2 antibody (C-18), and Bcr (breakpoint cluster region protein) were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-Gab1 antibodies and the anti-phosphotyrosine antibody 4G10 were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). Antibodies against HisG and Xpress tag were from Invitrogen. Antibodies that specifically recognize phospho-ERK1/2 were from New England Biolabs (Beverly, MA). Normal mouse and rabbit IgG and anti-FLAG M2 antibodies were obtained from Sigma.
Immunoprecipitation and Western Blot Analysis-After treatment with reagents, the cells were washed with ice-cold PBS (Ϫ), harvested in 0.5 ml of lysis buffer (50 mM sodium pyrophosphate, 50 mM NaF, 50 mM NaCl, 5 mM EDTA, 5 mM EGTA, 100 M Na 3 VO 4 , 10 mM HEPES, pH 7.4, 0.1% Triton X-100, 500 M phenylmethanesulfonyl fluoride, and 10 g/ml leupeptin). After a flash-freezing on liquid nitrogen and thawing on ice, cells were scraped off the dish and centrifuged at 14,000 ϫ g (40°C for 30 min), and protein concentration was determined using the Bradford protein assay (Bio-Rad). For immunoprecipitation, cell lysates were incubated with anti-HA, SHP-2, Bcr, Gab1, or Xpress antibody for 16 h at 40°C as indicated and then incubated with 20 l of protein Aor protein G-Sepharose CL-4B (Amersham Biosciences) for 1 h on a roller system at 40°C. The beads were washed three times with 500 l  of lysis buffer. For Western blot analysis, cell lysates or immunoprecipitates were subjected to SDS-PAGE, and proteins were transferred to nitrocellulose membranes (Hybond TM ECL; Amersham Biosciences). The membrane was blocked for 1 h at room temperature with 3% bovine serum albumin (ICN Biomedicals Inc.) blocking buffer. The blots were incubated for 4 h at room temperature with a primary antibody and then were followed by incubation for 1 h with horseradish peroxidaselinked secondary antibody (Amersham Biosciences). Immunoreactive bands were visualized using ECL (Amersham Biosciences).
In Vitro Phosphatase Activity Assay for SHP-2 Tyrosine Phosphatase Activity-SHP-2 phosphatase activity was assayed by using tyrosinephosphorylated Raytide as a substrate as previously described (21). In brief, cell lysates were prepared in lysis buffer (25 mM Tris-HCl, 10 mM 2-mercaptoethanol, 2 mM EDTA, 1 mM phenylmethanesulfonyl fluoride, 10 g/ml each leupeptin and aprotinin, 5 mM benzamidine, and 1% Nonidet P-40). The lysates were precleared with protein A/G-conjugated agarose beads and control IgG from normal rabbit serum, and SHP-2 and Xpress-tagged SHP-2 were immunoprecipitated by anti-SHP-2 or anti-Xpress antibody, respectively. After having been washed, the immune complexes were added to 50 l of 32 P-labeled Tyr-Raytide and 50 l of 2ϫ phosphatase reaction buffer and incubated at 30°C for 5 min. The reaction was terminated by the addition of acidic charcoal mixture (0.9 M HCl, 10 mM sodium pyrophosphate, 2 mM NaH 2 PO 4 , and 4% (w/v) NoritA). After centrifugation, the amount of radioactivity present in the supernatant was counted by a scintillation counter.
PathDetect Trans-reporting System-A PathDetect trans-system (Stratagene) was used for detection of Elk1 transcription activity as described previously (22). Statistical Analysis-Data are reported as mean Ϯ S.D. Statistical analysis was performed with the StatView 4.0 package (ABACUS Concepts, Berkeley, California). Differences were analyzed with unpaired two-tailed Student's t test or Welch's t test as appropriate. p values less than 0.05 are indicated by an asterisk, and p values less than 0.01 are indicated by a double asterisk.

Expression of a Catalytically Inactive Form of SHP-2 Inhibits ERK2 Activation in
Response to PDGF-We tested the feasibility of using PDGF-␤ receptor overexpressing CHO cells to study the role of SHP-2 in activation of ERK1/2 by PDGF, because CHO cells can be easily transfected to transiently express protein. cDNAs encoding wild type (WT) or catalytically inactive SHP-2 with a Cys to Ser mutation at residue 459 (C/S) were co-transfected into cells along with an HA epitopetagged ERK2. Cells were stimulated with 20 ng/ml PDGF-BB for the indicated times and lysed, HA-ERK2 was immunoprecipitated using anti-HA mAb, and ERK2 activity in the immunoprecipitates was determined by using anti-phosphospecific ERK1/2 antibody. As shown in Fig. 2, A and B, expression of catalytically inactive SHP-2 (SHP-2C/S) in PDGF-␤ receptoroverexpressing CHO cells significantly inhibited PDGF-induced ERK2 activation.
It has been reported that after the binding of the SHP-2 SH2 domains to a tyrosine-phosphorylated target such as Gab1, SHP-2 adopts an "open" conformation associated with increased catalytic activity. Therefore, we also determined the time course of PDGF-induced SHP-2 tyrosine phosphatase activity (Fig. 2C). As shown in Fig. 2C, PDGF (20 ng/ml) increased SHP-2 tyrosine phosphatase activity ϳ2.5-fold, which reached a peak at 10 min and returned to the basal level within 30 min.
Catalytic Inactive Form of SHP-2 (SHP-2C/S) Constitutively Associates with Gab1-Since SHP-2 and Gab1 interaction is critical for ERK1/2 activation and SHP-2C/S inhibits EGFinduced ERK activation, we evaluated SHP-2C/S association with Gab1. As shown in Fig. 2, D and E, cells transfected with SHP-2C/S had elevated levels of Gab1 tyrosine phosphorylation and Gab1-SHP-2 complex in the serum-starved condition. After PDGF stimulation, there was a continuous rise in Gab1 tyrosine phosphorylation and continuous increase in the amount of SHP-2 associated with Gab1 as previously reported (Fig. 2, D and E). These data suggest that SHP-2C/S can trap Gab1 in a tyrosine-phosphorylated state. It has been reported that SHP-2-Gab1 complex formation is critical for ERK1/2 activation (16). In regard to the regulation of ERK1/2 activation by SHP-2-Gab1 complex formation, the finding that SHP-2C/S constitutively associates with Gab1 is incompatible with the inhibitory effect of SHP-2C/S on PDGF-induced ERK1/2 and Elk-1 activation. Therefore, we investigated further the role of Gab1/SHP-2 interaction and SHP-2 phosphatase activity in SHP-2 signaling.
Since SHP-2-(399 -593) increased and SHP-2-(399 -593C/S) decreased wild type SHP-2 tyrosine phosphatase activity, we evaluated whether SHP-2-(399 -593) can associate with the C-terminal region of SHP-2 directly as a dimer. We co-trans-  (23). We also found that SHP-2-(1-202) and -(1-401), which contain the N-terminal SH2 domain, were able to form a complex with Gab1 by coimmunoprecipitation study with anti-Gab1 antibody (Fig. 5, A and B). We determined that the deletion mutant SHP-2-(198 -593), in which the N-terminal SH2 domain and a significant part of the C-terminal SH2 domain of SHP-2 is deleted, did not associate with Gab1 as previously reported (23). However, Gab1 was able to associate with the smaller construct SHP-2-(399 -593), com-  1 and 2 in A-C) or rabbit IgG as a control (lanes 3 and 4 in A and B). Immunoprecipitates (IP) (lanes 1-4 in A and B and lanes 1 and 2 in C) and total cell lysates (TCL) (lanes 5 and 6 in A and B and lanes 3 and 4 in C) were analyzed by immunoblotting (IB) with anti-Xpress (A and B) or anti-FLAG (C) antibody. No difference in the amount of immunoprecipitated Gab1 was detected by Western blot analysis with anti-Gab1 antibody (A and B (bottom)). posed of C-terminal SHP-2 (Fig. 5A). Interestingly, SHP-2-(399 -593C/S) also can associate with Gab1, and no significant difference between the association of Gab1 with SHP-2-(399 -593) and SHP-2-(399 -593C/S) was observed (Fig. 5C). Equal amounts of Gab1 protein were present in Gab1 immunoprecipitates (data not shown). These results indicate that Gab1 is able to interact with the C-terminal region of SHP-2 as well as with SHP-2 SH2 domains and that the mutation of Cys 459 to Ser did not affect Gab1 interaction with the C-terminal region of SHP-2.
Gab1 Can Interact with the C-terminal Region of the SHP-2 Tyrosine Phosphatase Domain-As shown in Fig. 5A, Gab1 was able to associate with SHP-2-(399 -593) but not with SHP-2-(198 -593). These data suggest that the domain of amino acids 198 -398 of SHP-2 has an inhibitory effect on Gab1/SHP-2-(399 -593) association. However, it is also possible that Gab1/ SHP-2-(399 -593) interaction is artificial, caused by the exposure of new surfaces by the deletion of aa 198 -398. The usual way to determine the association of Gab1/SHP-2 is to mutate the binding site for Gab1 in SHP-2-(399 -593). However, since the binding site of Gab1 is in the SHP-2 phosphatase domain, the mutation in SHP-2-(399 -593) may change the SHP-2 phosphatase activity and also the protein structure of SHP-2. Therefore, to determine whether Gab1 is able to interact with C-terminal region of SHP-2-(399 -593) in wild type SHP-2, we performed a competitive assay to evaluate Gab1 association with SHP-2-(399 -593) as follows.
As we stated in the Introduction, there are three possible models to explain the role of SHP-2 in Elk-1 activation: 1) the combination of both SHP-2-Gab1 complex formation and SHP-2 phosphatase activity is critical for ERK1/2 and Elk-1 activation; 2) Gab1 associates with N-terminal SH2 domains of SHP-2, increases SHP-2 phosphatase activity, and then activates ERK1/2 and Elk-1 activity; or 3) the association of Gab1 with SHP-2 is the only critical mechanism to regulate ERK1/2 and Elk-1 activity, and SHP-2 phosphatase activity is not directly involved in this process. Model 2 can explain the first Step 2). Both Gab1 interaction and the increasing counterpart of SHP-2 wild type phosphatase activity are necessary for Elk-1 activation. B, SHP-2C/S and SHP-2-(399 -593C/S) can form a dimer and bind with Gab1. However, these mutants do not have phosphatase activity. Furthermore, these mutants inhibit a counterpart of SHP-2 phosphatase activity and PDGF-induced Elk-1 activity. C, SHP-2-(198 -593) has SHP-2 phosphatase activity but is unable to associate with Gab1 and inhibit PDGF-induced Elk-1 activation.

TABLE I
Summary of the effects of SHP-2 mutants step of SHP-2 activation in Fig. 9A but not the inhibitory effect of SHP-2-(198 -593), which is a constitutively active form of SHP-2, on Elk-1 activation (Fig. 9C). In contrast, it is possible to explain the inhibitory effect of SHP-2-(198 -593) (Fig. 9C) on Elk-1 activation by model 3. However, the inhibitory effect of SHP-2C/S and SHP-2-(399 -593C/S) cannot be explained by model 3. If "trapping" of Gab1 by SHP-2C/S is the main inhibitory mechanism of SHP-2C/S as explained in the Introduction, it is difficult to explain how SHP-2-(399 -593) can activate Elk-1, because SHP-2-(399 -593) has no phosphatase activity but can associate with Gab1 similarly to SHP-2C/S. Therefore, model 1 is the most likely mechanism of SHP-2 function in Elk-1 activation shown in Fig. 9A.