PSTPIP Is a Substrate of PTP-PEST and Serves as a Scaffold Guiding PTP-PEST Toward a Specific Dephosphorylation of WASP*

PSTPIP is a tyrosine-phosphorylated protein involved in the organization of the cytoskeleton. Its ectopic expression induces filipodial-like membrane extensions in NIH 3T3 cells. We previously observed a defect in cytokinesis and an increase in the tyrosine phosphorylation of PSTPIP in PTP-PEST-deficient fibroblasts. In this article, we demonstrate that PTP-PEST and PSTPIP are found in the same complexes in vivo and that they interact directly through the CTH domain of PTP-PEST and the coiled-coil domain of PSTPIP. We tested pathways that could regulate the tyrosine phosphorylation of PSTPIP. We found that the activation of the epidermal growth factor and platelet-derived growth factor receptors can induce PSTPIP phosphorylation. With the use of the PP2 inhibitor, we demonstrate that Src kinases are not involved in the epidermal growth factor-medi-ated phosphorylation of PSTPIP. Together with previous results, this suggests that c-Abl is the critical tyrosine kinase downstream of mutants TPCK-treated were chromatography (TLC) in the binding of PSTPIP. The association between PTP-PEST and PSTPIP was reconstituted in a 293-T cell system. We did not detect endogenous PTP-PTP-PEST

Within the protein-tyrosine phosphatase (PTP) 1 superfamily, PTP-PEST is a ubiquitously expressed cytoplasmic enzyme with the conserved catalytic domain located at the amino terminus of the protein (1). PTP-PEST also contains a long carboxyl-terminal tail carrying several signal transduction motifs. Among these motifs, two classical proline-rich stretches have been shown to interact with the SH3 domains of Grb2, p130 Cas , Hef1/Cas-L, Sin/Efs, and Csk (2)(3)(4)(5). In addition, a novel and non-classical polyproline region in PTP-PEST has been implicated in the binding of paxillin and Hic-5 (6 -9). Finally, the PTB domain of the Shc proteins has been reported to interact with a NPLH sequence of PTP-PEST in a phosphotyrosineindependent manner (10). A common theme among PTP-PESTassociated proteins is their implication in the actin cytoskeleton and in focal adhesion signaling, suggesting a regulatory role for PTP-PEST in these specialize cellular structures (11).
The identification of p130 Cas as a substrate of PTP-PEST using catalytically inactive substrate trapping mutants of the enzyme was a major step in the elucidation of some of the biological functions of this protein (3,5,12). Although the PTP domain of PTP-PEST is sufficient to recognize the p130 Cas phosphorylated on tyrosine residues in vitro, an additional level of specificity is mediated via a SH3 domain-dependent association in vivo (3,5). In PTP-PEST Ϫ/Ϫ cells, p130 Cas is hyperphosphorylated and the cells exhibit impaired migration on a fibronectin matrix, presumably due to their inability to turn over focal adhesions and actin stress fibers (13). Additionally, in Rat-1 fibroblast cell lines overexpressing PTP-PEST, p130 Cas was demonstrated to be constitutively hypophosphorylated. In these cells, motility defects were also observed and explained by the failure of these cells to form the p130 Cas ⅐CrkII complex, which is required for integrin-dependent migration (14). Recently, it has been suggested that c-Abl, Shc, Pyk2, and Fak may well be additional substrates of PTP-PEST (15)(16)(17).
PTP-PEST belongs to a subfamily of PTPs that includes PEP, PTP-HSCF, and BDP (3,18). These PTPs are grouped together because, in addition to the homology in their PTP domains, they also contain a similar amino acid motif at their complete carboxyl-terminal termed the CTH domain. Previously, PTP-HSCF, a protein restricted to hemapoietic stem cells, was shown to interact with the proline, serine, threonine-rich phosphatase interacting protein (PSTPIP) via its CTH domain (18). The discovery that WASP, the protein encoded by the gene mutated in the Wiskott-Aldrich syndrome, is a ligand for the SH3 domain of PSTPIP creates an intriguing link between PSTPIP and actin reorganization (19).
PSTPIP appears to be the mammalian homologue of the fission yeast Schizosaccharomyces pombe CDC15p, a protein involved in the organization of the cytokinetic cleavage furrow (20). It was reported that PSTPIP associates with cortical actin during most of the cell cycle and translocates at the cleavage furrow during cytokinesis (18). Although the function of PST-PIP in the cleavage furrow can be thought to be similar to the function of CDC15p in yeast, the role of PSTPIP at the cortical actin during most of the cell cycle is completely unknown. Several mammalian proteins sharing a similar domain organization as PSTPIP have recently been identified. CIP4, PAC-SIN1/SyndapinI, PACSIN2/Syndapin2, PACSIN3, and PST-PIP2/MAYP all have an amino terminus homologous to CDC15p (21)(22)(23)(24)(25)(26). The amino terminus of these proteins can further be subdivided into a FCH (Fer kinase and CIP4 homology) and a CDC15-like coiled-coil domain (25). Recently, the FCH domain of CIP4 was shown to interact with microtubules (27). Interestingly, all these PSTPIP-like proteins, with the exception of PSTPIP2/MAYP, have a carboxyl-terminal SH3 domain involved in the binding of WASP and/or N-WASP. Together, these proteins form a family of distantly related members that could share overlapping functions in actin-dependent biological events.
WASP, N-WASP, and WAVEs (also know as SCARs) proteins interact with and activate the Arp2/3 actin nucleation activity (28,29). These proteins are inactive at the resting state because of an intramolecular interaction involving the Arp2/3binding site (known as the VCA or WCA domain) and GTPasebinding domain (30). Studies to date have demonstrated that the binding of GTP-loaded CDC42 and phosphatidylinositol 4,5-biphosphate (PIP 2 ) to N-WASP and WASP is sufficient to release the inhibitory intramolecular interactions and activate these proteins (31)(32)(33). Furthermore, SH3 domain-containing proteins, such as Grb2, Nck, and WISH, can interact with a proline-rich region found on the WASP-like proteins and cooperate in their activation (34 -36). The activation of the WASP proteins lead to actin nucleation via the recruitment and activation of the Arp2/3 complex, and to the formation of filipodia (N-WASP) and lamelipodia (WAVE) (33). These events are important in the generation of cell polarity, directed cell migration, vesicule transport and endocytosis (37)(38)(39).
In PTP-PEST Ϫ/Ϫ cells, we previously noted the hyperphosphorylation of PSTPIP and a defect in cytokinesis (13). This guided us to investigate if PSTPIP can interact directly with PTP-PEST. In this article, we report that PSTPIP and PTP-PEST associates in vivo. The coiled-coil domain of PSTPIP and the CTH domain of PTP-PEST mediate this interaction. We also demonstrate that PTPIP becomes tyrosine phosphorylated in mouse fibroblasts following EGF or PDGF treatment. This phosphorylation of PSTPIP is independent of Src kinases since the PP2 inhibitor did not affect it. We provide evidence that PTP-PEST negatively regulates the association of PSTPIP with SH2 domain-containing proteins by the dephosphorylation of PSTPIP on tyrosine 344. In contrast, the association of WASP with PSTPIP was demonstrated to be constitutive in vivo and independent of PSTPIP tyrosine phosphorylation. However, PSTPIP creates a physical link between PTP-PEST and WASP resulting in a specific PTP-PEST-mediated dephosphorylation of WASP. Cell Culture and Transfections-NIH 3T3, HER14, HEK293-T,  COS-1, and RAW 264.7 cell lines were grown in Dulbecco's modified  Eagle's medium. 70Z/3 and BaF/3 pro B-cells (obtained from Dr. A. Veillette, IRCM, Montréal) were grown in RPMI 1640. All cell culture media were supplemented with 10% fetal bovine serum (BIOSOURCE) and a mixture of penicillin/streptomycin (Invitrogen). COS-1, NIH 3T3, and HER14 cell lines were serum starved for 48 h in media containing 0.1% serum. These cells were treated with either 25 ng/ml PDGF (Roche Molecular Biochemicals), 100 ng/ml EGF (Sigma), 1 M insulin (Invitrogen) or media alone for 15 min prior to lysis. The PP2 inhibitor (Calbiochem) was used at 10 M in Dulbecco's modified Eagle's medium and was added to the cells 30 min prior EGF stimulation. HEK293-T and COS-1 cells were transfected with 5-10 g of total plasmid DNA using the calcium phosphate precipitation method as previously described (40).

EXPERIMENTAL PROCEDURES
Antibodies-The polyclonal antibody against PSTPIP (number 2221) and its HRP conjugate were previously reported (13). The anti-PTP-PEST (number 1075) was described (40). Two additional polyclonal antibodies were generated in this study against PTP-PEST: number 2528 (immunogen: GST-344 -437) and number 2530 (immunogen: GST-471-613). The anti-Flag M2 monoclonal antibody was obtained from Sigma. The anti-c-Src, FAK, c-Cbl, c-Abl, and WASP were from Santa Cruz Biotechnologies. The anti-Fyn was a gift from Dr. A. Veillette. The anti-Lyn was purchased from Pharmingen.
Plasmids and Constructions-PTP-PEST, WT, and C231S, in the expression vector pcDNA3.1 Zeo(ϩ) were previously described. PTP-PEST with a deleted Pro1 or Pro2 (reported in Refs. 3 and 6) were subcloned from pACTAG into the NotI and XbaI sites of pcDNA3.1 Zeo. PTP-PEST with a deleted CTH domain (⌬CTH) was constructed by digesting the pcDNA3.1 Zeo PTP-PEST WT with PstI and religation of the vector following gel purification (QIAEX II, Qiagen). To construct pcDNA3.1 Zeo PTP-PEST C231S ⌬Pro3-5, the PTP-PEST C231S cDNA was digested with BamHI and ApaI and ligated in the same sites of pcDNA3.1 Zeo. To construct EFGP PTP-PEST, the cDNA of PTP-PEST was amplified by PCR using engineered oligonucleotides containing BamHI and SalI restriction sites, and the digested PCR product was ligated into pEGFP-C2 linearized with BglII and SalI. EGFP PTP-PEST ⌬CTH was generated by subcloning the BamHI and PstI fragment of the PTP-PEST cDNA into BglII and PstI sites of pEGFP-C1. Full-length PSTPIP in pRK5 Flag C was previously described (18). PSTPIP was subcloned in pcDNA3.1 Zeo in the NotI/XhoI sites. Tryptophan 232, tyrosine 344, and tyrosine 367 were mutated by strand overlap extension PCR using oligonucleotides with point mutations as described in Ref. 6. These procedures resulted in the following mutants: W232A (coiled-coil domain mutant), Y344F, and Y367F. All the mutations were verified by dideoxy sequencing using Sequenase (Amersham Bioscience, Inc.). The plasmid constructions for the GST fusion proteins of the SH2 domain of Crk, c-Abl, Fyn, and c-Src were obtained from Dr. M. Park (McGill University). The SH2 domain of Nck was a gift from Dr. L. Larose (McGill). pcDNA3.1 LCK, pSR␣ v-Src, pcDNA3.1 avian c-Src, and Fyn were kind gifts from Dr. A. Veillette. pME Lyn was a kind gift from Dr. T. Yamamoto (The University of Tokyo, Japan), c-Abl cDNA was a gift from Dr. O. N. Witte (University of California, Los Angeles, CA), and the cDNA of WASP was a kind gift of Dr. N.
Immunoprecipitation-Cells were lysed in 1 ml of lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 1% Nonidet P-40 supplemented with 1 mM sodium orthovanadate, 5 mM NaF and a Complete protease inhibitor mixture (Roche Molecular Biochemicals) for 10 min on ice. For in vivo substrate trapping, the lysis buffer was modified in its sodium orthovanadate concentration (0.1 mM instead of 1 mM) to avoid competition of the inhibitor with the C231S mutants of PTP-PEST. Following growth factor treatments, NIH 3T3 and HER14 cells were lysed in 1 ml of RIPA (40) for 10 min and diluted 1:1 with the Nonidet P-40 lysis buffer described above. The lysates were harvested and cleared by centrifugation at 16,000 ϫ g at 4°C. Aliquots of cell extracts (500 -1000 g of proteins) were immunoprecipitated with 1-4 l of polyclonal antibodies and 20 l of Protein-A agarose in a final volume of 1 ml for 90 min at 4°C. The immunoprecipitates were washed three times in 1 ml of lysis buffer, resuspended in 40 l of SDS sample buffer, and finally boiled for 5 min at 95°C. Typically, a fourth of the immunoprecipitations (10 l) were resolved by 7.5% SDS-PAGE and the proteins were transferred to PVDF membranes (Immobilon-P, Millipore).
In Vitro Translation-pACTAG PSTPIP was linearized with XbaI and used as a template for in vitro transcription using T7 RNA polymerase. Two microliters of the transcription reaction were used to per-form in vitro translation in the presence of 35 S-labeled methionine and rabbit reticulocyte lysate (Promega). The in vitro translated PSTPIP was assayed for binding to GST alone and GST PTP-PEST fusion proteins as described below.
In Vitro Binding Assay Using GST Fusion Proteins-GST fusion proteins were expressed in exponentially growing Escherichia coli DH5␣ by induction for 2 h with 1 mM isopropyl-1-thio-␤-D-galactopyranoside. Bacteria were harvested by centrifugation (typically 1.5 ml per binding assay) and the fusion proteins were isolated on gluthatione-Sepharose as previously described (3). Each fusion protein was incubated with the appropriate cell extract in immunoprecipitation buffer for 90 min at 4°C. The beads were washed three times with 1 ml of lysis buffer, resuspended in 40 l of SDS sample buffer, and boiled for 5 min at 95°C. The GST pull-downs were resolved by 7.5% SDS-PAGE and the proteins were transferred to PVDF membranes.
Metabolic Labeling and Phosphopeptide Maps-Experiments were performed as described in Refs. 41 and 42. Briefly, transfected HEK293-T cells were rinsed twice and incubated in phosphate-free Dulbecco's modified Eagle's medium (0.5% dialyzed fetal bovine serum) for 1 h. The cells were then incubated with 0.25 mCi/ml [ 32 P]orthophosphate in phosphate-free Dulbecco's modified Eagle's medium (0.5% dialyzed fetal bovine serum) for 4 h. The cells were rinsed twice with ice-cold 50 mM HEPES, pH 7.5, and lysed as described in the immunoprecipitation section. Cellular debris were removed by centrifugation and the PSTPIP proteins were immunoprecipitated with 3 l of the polyclonal antibody number 2221. The immunoprecipitated proteins were subjected to SDS-PAGE and transferred to PVDF. The membranes were air-dried and autoradiographed. The regions of PVDF containing PSTPIP or its mutants were cut and digested with 10 g of TPCK-treated trypsin for 16 h, followed by the addition of 5 g of TPCK-treated trypsin for 3 h. The samples were rinsed with water and lyophilized. This procedure was repeated three times. The samples were resuspended in electrophoresis buffer (2.5% (v/v) formic acid and 7.8% (v/v) glacial acetic acid) and spoted on cellulose-coated thin layer chromatography (TLC) plates. The electrophoresis was performed at pH 1.9 in a HTLE 7000 apparatus. The samples were separated in a second dimension by thin layer chromatography in TLC buffer (39% (v/v) 1-butanol, 30% (v/v) pyridine, 6.1% (v/v) glacial acetic acid). The TLC plates were dried and autoradiographed to visualize the phosphopeptides.
Western Blotting-PVDF membranes were blocked in TBS-T (20 mM Tris-HCl, pH 7.6, 137 mM NaCl, 0.1% Tween 20) containing either 1% bovine serum albumin or 5% nonfat dry milk for 60 min. The primary antibodies were incubated with the blot in blocking buffer for 60 min at room temperature or overnight at 4°C. The polyclonal anti-PSTPIP (number 2221), anti-PSTPIP-HRP (number 2221), and anti PTP-PEST (numbers 2528 and 2530) antibodies were used at 1:5000 dilutions in TBS-T with 5% milk. The monoclonal antibodies against phosphotyrosine, 4G10 and 4G10-HRP, were used at 1:5000 dilutions in TBS-T with 1% bovine serum albumin. The commercial antibodies were diluted according to the manufacturers recommendations. The membranes were washed two times for 5 min and two times for 10 min with TBS-T and incubated when needed with the appropriate secondary antibody conjugated to horseradish peroxidase for 45 min. The blots were washed extensively by repeated incubation in TBS-T and the bound antibodies were detected using chemiluminescence (Renaissance, PerkinElmer Life Sciences).
Immunofluorescence-COS-1 cells were transfected to transiently co-express EGFP PTP-PEST (WT or ⌬CTH) and PSTPIP. Cells were serum starved overnight, fixed with 4% (w/v) of paraformaldehyde, permeabilized, and blocked as described in Ref. 13. The anti-PSTPIP antibody (number 2221) was used at a 1:1000 dilution in PBS containing 2% bovine serum albumin. The anti-rabbit antibody conjugated to TRITC (Jackson ImmunoResearch Laboratories Inc.) was used at a 1:200 dilution. Cells were visualized with a Nikon Eclipse TE300 microscope using a ϫ40 objective (Nikon). Pictures were acquired employing a digital camera (model Quantix, Photometrics) and the MetaMorph software (Universal Imaging Corp.).

PSTPIP and PTP-PEST Expression Overlaps in Mouse
Tissues and Cell Lines-We previously reported a defect in cytokinesis in addition to the hyperphosphorylation of PSTPIP in PTP-PEST Ϫ/Ϫ cells (13). This guided us to investigate if PSTPIP can interact directly with PTP-PEST. A graphical representation of PSTPIP and its mutant used in this study is presented in Fig. 1A. PSTPIP contains an amino-terminal FCH domain, a coiled-coil domain, a SH3 domain, and two putative tyrosine phosphorylation sites. We generated a polyclonal antibody against the SH3 domain of PSTPIP (Fig. 1A). This domain of PSTPIP has been reported to become tyrosine phosphorylated (19). Since several experiments using this antibody will aim at characterizing the tyrosine phosphorylation level of PSTPIP and its association with other proteins, we studied the anti-PSTPIP antibody in more detail. As seen in Fig. 1B, the antibody can precipitate PSTPIP from NIH 3T3 fibroblasts and 70Z/3 pro-B cells (bottom panel), either unphosphorylated or highly tyrosine phosphorylated because of pervanadate treatment of the cells prior to lysis (upper panel). In addition, the antibody could co-precipitate several tyrosine phosphorylated proteins together with PSTPIP from pervanadate-treated NIH 3T3 and 70Z/3 cell lysates (upper panel, lanes 4 and 8). The preimmune serum did not precipitate any PSTPIP nor any other tyrosine-phosphorylated protein.
Little is known concerning the expression pattern of PSTPIP at the protein level. Before investigating the interactions between PTP-PEST and PSTPIP, we examined if the proteins were significantly co-expressed in a panel of mouse tissues. As seen in Fig. 1C, PSTPIP was detected in high amounts in spleen and thymus, followed by a more moderate expression in lung, brain, and muscle lysates. Low levels of PSTPIP were detected in heart and liver and no protein was detected in kidney extracts. We do not know at the present if the faster migrating band observed in heart and liver lysates corresponds to a PSTPIP-like protein or a degradation product. PTP-PEST expression significantly overlaps with the pattern of PSTPIP at the protein level (Fig. 1C). In particular, the two proteins are highly present in spleen, thymus, lung, and brain and both proteins are absent from kidney lysates. Because PSTPIP was most highly found in spleen and thymus, we compared the expression of PTP-PEST and PSTPIP in a panel of hematopoietic cells. PSTPIP was highly expressed in two pro-B cell lines, 70Z/3 and BaF/3, and in RAW macrophages. Lower levels of PSTPIP were found in two T-cell lines, Jurkat and H9 (not shown). In this set of cell lines, PTP-PEST was detected at high level in all samples. In conclusion, the patterns of expression of PSTPIP and PTP-PEST are very similar and this suggests that the study of their interaction might be of biological significance.
PSTPIP Is Associated with PTP-PEST in Fibroblasts and Hematopoietic Cells-We investigated if PSTPIP interacts with PTP-PEST in vivo by a co-immunoprecipitation experiment. BaF/3 pro-B cells and NIH 3T3 fibroblasts were lysed and PSTPIP was immunoprecipitated with an anti-PSTPIP antibody. Blotting with a polyclonal antibody against PTP-PEST (number 2530) was performed to monitor the presence of the co-precipitated PTP-PEST. As seen in Fig. 2A, PTP-PEST was detected in PSTPIP immunoprecipitates both from BaF/3 and NIH 3T3 (top panel, lanes 2 and 4). As a control, no PTP-PEST was found when the preimmune serum was used (lanes 1 and 3). The blot was reprobed with an anti-PSTPIP-HRP antibody to verify the precipitation of PSTPIP (bottom panel). In addition, we also observed the association between the two proteins when PTP-PEST was immunoprecipitated (not shown). Thus, PSTPIP and PTP-PEST are found together in a complex in vivo both in fibroblasts and hematopoietic cells.
The CTH Domain of PTP-PEST Is Required to Bind PST-PIP-We next wanted to characterize the requirement for the association between PTP-PEST and PSTPIP. The carboxyl end of PTP-HSCF is highly homologous to PTP-PEST carboxyl terminus. This region of PTP-HSCF was termed the CTH domain and was shown to be involved in the binding of PSTPIP. The association between PTP-PEST and PSTPIP was reconstituted in a 293-T cell system. We did not detect endogenous PTP-PEST and PSTPIP in this cell line with our antibodies. PTP-PEST constructs (WT, ⌬Pro1, ⌬Pro2, and ⌬CTH) were transiently co-transfected with PSTPIP. PTP-PEST was immunoprecipitated from each lysate (including mock and PSTPIP alone controls) and the co-precipitation of PSTPIP was analyzed by Western blotting with the anti-PSTPIP-HRP antibody. As observed in Fig. 2B (middle panel), PSTPIP was precipitated with WT, ⌬Pro1, and ⌬Pro2 PTP-PEST but was completely absent from a precipitation of PTP-PEST ⌬CTH. The precipitation of the various PTP-PEST proteins was verified by Western blotting (Fig. 2B, bottom panel). The expression of PSTPIP was ascertained by Western blotting the total cell lysate with the anti-PSTPIP antibody (Fig. 2B, top panel). In support of these data, we observed that the isolated CTH domain of PTP-PEST expressed as a GST fusion protein was sufficient to interact with [ 35 S]PSTPIP in a pull-down assay as seen in Fig. 2C. We conclude that the PTP-PEST CTH domain is responsible for its direct interaction with PSTPIP.
The Coiled-coil Domain of PSTPIP Is Responsible for the Interaction with PTP-PEST-To investigate, in vivo, the functional domain of PSTPIP involved in the binding of PTP-PEST, we reconstituted the association between PTP-PEST and PST-PIP (WT, W232A, Y344F, and Y367F) in 293-T cells. We mutated a critical tryptophan residue (W232A) in the PSTPIP coiled-coil domain known to be essential for the binding to PTP-HSCF (43). As shown in Fig. 2D (middle panel), PTP-PEST co-precipitated with WT, Y344A, and Y367A PSTPIP but not with PSTPIP W232A. The blot was stripped and reprobed with an anti-PSTPIP-HRP antibody (Fig. 2D, bottom panel). The expression level of PTP-PEST was assayed by Western blot analysis of the lysates (Fig. 2D, top panel). Furthermore, we observed that a GST fusion of the SH3 domain of PSTPIP was not able to precipitate PTP-PEST. However, the GST fusion of the full-length PSTPIP did interact strongly with HA-PTP-PEST suggesting the involvement of the amino terminus of PSTPIP in this process as seen in Fig. 2E.
PTP-PEST and PSTPIP Co-localize in COS-1 Cells-To support our biochemical interaction data, we investigated if PTP-PEST and PSTPIP could co-localize in COS-1 cells by immunofluorescence microscopy. The cells were transiently transfected with plasmids encoding PSTPIP and enhanced green fluorescent protein (EGFP)-tagged PTP-PEST (WT or ⌬CTH). As seen in Fig. 3, A and B, the PTP-PEST ⌬CTH mutant was detected as a diffuse signal throughout the cytoplasm and the co-transfected PSTPIP gave a fibrous staining and was also detected at the cortex of the stained cells. The two proteins were not significantly co-localized as observed by analyzing the yellow signal in the images overlay (Fig. 3C). PTP-PEST WT was also detected as a diffuse signal throughout the cytoplasm but, interestingly, the co-expressed PSTPIP was not detected as a fibrous signal but rather gave a diffuse cytoplasmic staining in these conditions (Fig. 3, D and E). Thus, there is a clear difference in the localization of PSTPIP when it is expressed with either PTP-PEST WT or ⌬CTH. As observed in the images overlay (Fig. 3F), almost all of the PTP-PEST and PSTPIP were co-localized in the cytoplasm and the perinuclear region as judged by the yellow signal. We verified that the large GFP-tag added to PTP-PEST did not interfere with the ability of PTP-PEST to associate with PSTPIP. As seen in panel G, GFP PTP-PEST WT, but not ⌬CTH mutant, nicely co-precipitated with PSTPIP. In summary, we found that PTP-PEST and PST-PIP are co-localized in COS-1 cells and that the expression of WT but not ⌬CTH PTP-PEST results in a lost of the cortical and fibrous staining of PSTPIP. A, schematic representation of PSTPIP and its mutants used in this study. B, characterization of the antibody against PSTPIP. NIH 3T3 and 70Z/3 were treated or not with pervanadate and PSTPIP was immunoprecipitated from the cell lysate with either the preimmune or the anti-PSTPIP serum to ascertain that tyrosine phosphorylation of PSTPIP does not affect recognition by the antibody. PSTPIP was precipitated to the same extent in both cell lines (bottom panel) and regardless of its phosphorylation state (top panel). No signal was detected in the preimmune precipitations demonstrating the specificity of the antibody. C, Western blotting was performed on 25 g of protein extracts from various mouse tissues using antibodies against PSTPIP and PTP-PEST.

PSTPIP Is Tyrosine Phosphorylated Upon PDGF and EGF Treatments in NIH 3T3
Fibroblasts-The pathways involved in inducing PSTPIP tyrosine phosphorylation in vivo have not been fully studied. In a previous report, activation of the PDGF receptor resulted in the tyrosine phosphorylation of PSTPIP (15). This event was dependent on c-Abl since in Abl Ϫ/Ϫ fibroblasts PSTPIP failed to become tyrosine phosphorylated when the cells were treated with PDGF. We investigated whether EGF, PDGF, and insulin could affect PSTPIP tyrosine phosphorylation in mouse fibroblasts. NIH 3T3 and NIH 3T3 expressing the human EGF receptor (HER14) cells were serum starved and treated with media alone, 25 ng/ml PDGF, 1 M insulin or pervanadate for 15 min. HER14 cells were stimulated with 100 ng/ml EGF for 10 min. Each cell lysate was subjected to either a preimmune or an anti-PSTPIP immunoprecipitation. As observed in Fig. 4A, treatment of NIH 3T3 cells with PDGF results in a 2.4-fold increase in PSTPIP tyrosine phosphorylation levels (as determined by densitometry analysis) while insulin had no effect (1.2-fold increase). Treatment of HER14 cells with 100 ng/ml EGF stimulated the tyrosine phosphorylation of PSTPIP by 2.3-fold when compared with the serum-starved control. Finally, treatments of HER14 cell with pervanadate lead to a dramatic increase (4.5-fold) in PSTPIP tyrosine phosphorylation. The blot was reprobed with the anti-PSTPIP-HRP to ascertained PSTPIP precipitation (Fig. 4A, lower panel).
We screened a panel of tyrosine kinases involved in PDGF and EGF receptor signaling to obtain insight on the mechanism involved in PSTPIP phosphorylation. We investigated if members of the Src kinases (v-Src, c-Src, Fyn, Lck, and Lyn), c-Abl or FAK are individually able to phosphorylate PSTPIP following transient co-expression in 293-T cells. PSTPIP was immunoprecipitated and its phosphotyrosine levels were analyzed. As observed in Fig. 4B (top panel), PSTPIP was tyrosine phosphorylated when co-transfected with v-Src, c-Src, Fyn, Lyn, and c-Abl. The hematopoietic restricted tyrosine kinase Lck and the focal adhesion kinase (FAK) were not able to phosphorylate PSTPIP in this assay. The amount of precipitated PST-PIP was evaluated by Western blotting with the anti-PSTPIP-HRP antibody as seen in Fig. 4B (middle panel). With the exception of Lck, we verified the expression of the tyrosine kinases by Western blotting (bottom panel). In conclusion, two widely distributed members of the Src family, c-Src and Fyn, and c-Abl are candidates to mediate the PDGFR and EGFR We next used the PP2 inhibitor to determine if Src kinases are involved in the tyrosine phosphorylation of PSTPIP downstream of the EGF receptor. The pretreatment of HER14 cells with PP2 did not block the EGF-induced tyrosine phosphorylation of PSTPIP as shown in Fig. 4C (top panel). The blot was reprobed with the anti-PSTPIP-HRP antibody to verify equal precipitation (Fig. 4C, bottom panel). We also observed that EGF stimulation does not affect the interaction between PTP-PEST and PSTPIP (Fig. 4C, middle panel). The effect on the PP2 inhibitor was ascertained by analyzing the total tyrosine phosphorylation in the cell lysate (Fig. 4D) and also by looking at the amount of c-Cbl, a known Src substrate, that was precipitated by the GST Fyn SH2 domain (Fig. 4E). In both control experiments, we indeed observed that the PP2 inhibitor blocked the activity of the Src kinases. We conclude that Src kinases are not involved in the phosphorylation of PSTPIP downstream of the EGF receptor and this result together with previous results (15) points to c-Abl as the critical kinase involved in PSTPIP phosphorylation downstream of growth factor receptors.
PTP-PEST Dephosphorylates Tyrosine 344 of PSTPIP and Prevents Its Interaction with SH2 Domain-containing Proteins-We next investigated if PSTPIP is a substrate of PTP-PEST and if the binding between the two proteins is required for an efficient dephosphorylation. PSTPIP and c-Abl were co-transfected with PTP-PEST WT, PTP-PEST ⌬CTH, PTP-PEST C231S, or PTP-PEST ⌬Pro3-5 C231S. The various PTP-PEST proteins were immunoprecipitated with an antibody recognizing PTP-PEST and all its mutants (serum number 2528). As observed in Fig. 5A (top panel), tyrosine-phosphorylated PSTPIP was precipitated by the C231S mutant of PTP-PEST. The blot was reprobed with an antibody against PSTPIP to demonstrate that the phosphorylated protein of 50 kDa is PST-PIP (middle panel). PSTPIP precipitated with WT PTP-PEST (lane 5) was dephosphorylated demonstrating that PSTPIP is a substrate for the catalytic activity of PTP-PEST. When c-Abl was not co-transfected, no tyrosine-phosphorylated PSTPIP could be precipitated by either PTP-PEST WT or C231S (lane 1  and 3). To establish if the CTH domain of PTP-PEST plays a role in the substrate recognition, we used a substrate trapping mutant (C231S) with a large deletion of the PSTPIP-binding site (⌬Pro3-5). As observed in Fig. 5A (top panel, lane 8), this mutant of PTP-PEST failed to precipitate any phosphorylated protein of 50 kDa, clearly demonstrating that the catalytic domain of PTP-PEST is not capable of interacting with tyrosine-phosphorylated PSTPIP on its own. All the PTP-PEST constructs in which the CTH domain is lacking did not interact with PSTPIP (middle panel). The expression of PTP-PEST proteins was verified by immunoblotting the total cell lysates with a polyclonal antibody against PTP-PEST (bottom panel).
We used tyrosine to phenylalanine mutants of PSTPIP (Y344F and Y367F) to investigate which tyrosine(s) of PSTPIP are dephosphorylated by PTP-PEST. c-Abl, PSTPIP, mutants, and PTP-PEST (WT or C231S) were co-transfected in 293-T cells and PTP-PEST was immunoprecipitated from each sample using the anti-PTP-PEST antibody (number 2530). As shown in Fig. 5B (top panel), PSTPIP Y344F was not detected in a tyrosine-phosphorylated form when co-immunoprecipitated with PTP-PEST C231S trapping mutant. However, PST-PIP Y367F was detected as a highly tyrosine-phosphorylated protein in the PTP-PEST C231S precipitate (lanes 2 and 6) but was almost completely dephosphorylated in the PTP-PEST WT precipitate (lanes 1 and 5). The blot was reprobed with an anti-PSTPIP antibody to verify that PSTPIP and the mutants were co-precipitated with the PTP-PEST proteins (Fig. 5B, middle panel). The lysates were immunoblotted with an anti-PTP-PEST antibody to monitor the expression of PTP-PEST (bottom panel). These results indicate that tyrosine 344 of PSTPIP is a target of PTP-PEST catalytic activity. No conclusion could be made about the action of PTP-PEST on the tyrosine 367 of PSTPIP in this assay since this tyrosine failed to become tyrosine phosphorylated when PSTPIP Y344F mutant was used with c-Abl or any other kinase that was tested. This issue is addressed in Fig. 7 of this article. From these data, we conclude that PTP-PEST dephosphorylates tyrosine 344 of PSTPIP.
We investigated if the phosphotyrosines of PSTPIP could serve as docking sites for SH2 domain containing proteins. To verify that hypothesis, we transfected 293-T cells with PSTPIP in combination with c-Abl to induce its tyrosine phosphorylation. In addition, since two putative tyrosine phosphorylation sites are reported (19), the PSTPIP mutants Y344F and Y367F were also co-transfected with c-Abl. Equal amounts of the cell lysates were subjected to pull-down assays with the SH2 domains of Crk, Fyn, c-Abl, Nck, or c-Src expressed as GST fusion proteins. As observed in Fig. 5C, the SH2 domains of c-Src, Fyn, and Abl precipitated PSTPIP. In contrast, the SH2 domain of Crk precipitated a very small amount of PSTPIP. GST Nck SH2 domain and GST alone did not precipitate PSTPIP. All the SH2 domains and GST alone did not precipitate PSTPIP when c-Abl was not co-transfected (not shown). We investigated if the SH2 domains of Fyn, c-Src, and c-Abl interacted

FIG. 3. PTP-PEST and PSTPIP co-localize in COS-1 cells. COS-1 cells were transfected with PSTPIP in combination with either EGFP PTP-PEST ⌬CTH (A and B) or EGFP PTP-PEST wild-type (D and E).
The EGFP fusion proteins signal is found in panels A and D while the TRITC signal (PSTPIP) is shown in panels B and E. Panels C and F correspond to the overlay of the EGFP and TRITC signals. All fields were photographed using a ϫ40 Nikon objective. G, EGFP PTP-PEST associates with PSTPIP. PSTPIP was co-transfected with either EGFP PTP-PEST WT or ⌬CTH. PSTPIP was immunoprecipitated and the bound PTP-PEST was detected by Western blotting. The precipitation of PSTPIP was verified by reprobing with the PSTPIP antibody. The expression of each proteins was also verified by immunoblotting the cell lysates.
preferentially with any of the two putative phosphotyrosines of PSTPIP (Tyr 344 and Tyr 367 ). PSTPIP Y344F was unable to interact with any of the SH2 domain tested (Fig. 5C, right   panel). In fact, PSTPIP phosphorylation is completely lost when tyrosine 344 is mutated (19). PSTPIP Y367F was effectively precipitated by the same SH2 domains as the wild type  ). B, the same experiment as in A was repeated with PSTPIP mutants Y344F and Y367F. C, PSTPIP interacts with SH2 domains in vitro. The SH2 domains of Crk, Fyn, c-Abl, Nck, and c-Src expressed as GST fusion protein, namely the SH2 domains of Fyn, c-Abl, and c-Src. In addition, the amount of PSTPIP Y367F precipitated by the SH2 domains was comparable with the amount of wild type PSTPIP precipitated by the same SH2 domains suggesting that Tyr 344 is the major site involved in these interactions.
Having established that tyrosine 344 is a binding site for SH2 domain-containing proteins, we investigated if PTP-PEST could regulate these associations by dephosphorylating tyrosine 344 of PSTPIP. PSTPIP Y367F and c-Abl were co-transfected with or without PTP-PEST. Each cell lysate was subjected to a precipitation with GST alone, GST SH2 c-Abl, or GST SH2 Fyn. As presented in Fig. 5D, the SH2 domain of c-Abl and Fyn precipitated a significant amount of PSTPIP Y367F but this binding was lost when PTP-PEST was coexpressed (top panel). The integrity of the GST fusion was verified by Coomassie staining. The expression of PTP-PEST and PSTPIP Y367F was ascertained by immunoblotting the lysates with the appropriate antibodies. In summary, PTP-PEST is able to dephosphorylate PSTPIP at tyrosine 344 and this event negatively regulates the association of PSTPIP with SH2 domain-containing proteins.
The Formation of the PSTPIP-WASP Complex Is Not Regulated by Tyrosine Phosphorylation in Vivo-It was previously reported that the phosphorylation of tyrosine 367, which lies within the SH3 domain of PSTPIP, negatively regulates the binding of PSTPIP with WASP in vitro (19). We tested if this negative regulation also occurs in vivo. WASP and PSTPIP were co-transfected with the tyrosine kinases c-Abl and Fyn, and we subsequently verified the stability of the PSTPIP⅐WASP complex. Surprisingly, even if PSTPIP tyrosine phosphorylation levels were dramatically increased under these conditions ( Fig. 4B and data not shown), the formation of the PSTPIP⅐WASP complex was not affected as observed in Fig. 6A (middle panel). The expression of WASP was detected in each lysate and similar amount of PSTPIP were immunoprecipitated as shown in the upper and lower panel of Fig. 6A.
It was originally reported that v-Src can induce the phosphorylation of tyrosine 367 of PSTPIP, at least by monitoring the formation of a slower migrating form of PSTPIP by SDS-PAGE (15,19). We investigated if this viral form of Src is capable of abrogating the PSTPIP⅐WASP complex in vivo. As observed in Fig. 6B, WASP does not co-immunoprecipitate with PSTPIP when v-Src is co-expressed (lanes 1 and 2). The co-expression of PTP-PEST was not sufficient to restore the binding between PSTPIP and WASP under these conditions (lane 3). Surprisingly, the co-expression of PTP-PEST only resulted in a 2-fold decrease in the phosphorylation of PSTPIP (middle panel). Interestingly, we observed that in the samples transfected with v-Src, the migration of PTP-PEST on SDS-PAGE was significantly retarded (lane 3, PTP-PEST blot). We failed to detect tyrosine phosphorylation of PTP-PEST under these conditions to explain this slower migration in SDS-PAGE. It is an intriguing possibility that v-Src expression could affect PTP-PEST catalytic activity. It is possible that a v-Src-activated serine kinase phosphorylates and down-regulates PTP-PEST catalytic activity since serine phosphorylation was proposed to regulate PTP-PEST activity (44). To demonstrate if the loss of association between PSTPIP and WASP results from the phosproteins were tested for their ability to interact with tyrosine-phosphorylated PSTPIP. Abl was co-transfected with WT, Y344F, or Y367F PSTPIP and the lysates were subjected to a binding assay using various GST SH2 domains. D, PTP-PEST regulates the binding between PSTPIP and SH2 domain containing proteins by dephosphorylating tyrosine 344. PSTPIP Y367F and c-Abl were co-transfected with or without PTP-PEST. Each cell lysate was subjected to a precipitation by GST alone, GST SH2 Abl or GST SH2 Fyn. The presence of the precipitated PSTPIP was monitored by immunoblotting with the anti-PSTPIP-HRP antibody. The presence and integrity of each fusion protein was verified by Coomassie Blue staining of the membrane.
FIG. 6. The PSTPIP⅐WASP complex is stable in vivo but can be disrupted by v-Src. A, PSTPIP and Flag-tagged WASP were co-transfected with an empty vector, c-Abl, or Fyn encoding plasmids. The level of Flag WASP in each transfection was verified by blotting the lysates with an antibody against the Flag epitope (upper panel). PSTPIP was immunoprecipitated and the co-precipitation of WASP was assayed by blotting with the anti-Flag antibody (middle panel). The immunoprecipitation of PSTPIP was verified by reprobing the blot with an anti-PSTPIP antibody (bottom panel). B, in a similar assay, v-Src ability to interfere with the PSTPIP⅐WASP complex was tested. PSTPIP and WASP were transfected with v-Src and/or PTP-PEST. PST-PIP was immunoprecipitated and Western blotting with the anti-Flag antibody monitored in the presence of co-precipitated WASP. The blot was sequentially reprobed with an anti-4G10 and an anti-PSTPIP antibody. The expression of WASP and PTP-PEST was also verified by immunoblotting the lysates with the anti-Flag and anti-PTP-PEST antibodies. C, the dissociation of PSTPIP and WASP mediated by v-Src is independent of PST-PIP tyrosine phosphorylation. PSTPIP Y367F mutant was transfected with either an empty vector or with v-Src. PST-PIP Y367F was immunoprecipitated and the presence of WASP was assayed by blotting with the anti-Flag antibody.
phorylation of tyrosine 367, we repeated the experiment using PSTPIP Y367F mutant. As observed in Fig. 6C, v-Src was unexpectedly able to prevent the association of PSTPIP and WASP even in the absence of Tyr 367 in PSTPIP. These results suggest that the ability of v-Src to regulate the association of PSTPIP and WASP is not dependent on the phosphorylation of tyrosine 367 of PSTPIP.
PSTPIP Is Not Phosphorylated on Tyrosine 367-Experimental evidence, such as retarded migration in SDS-PAGE and loss of WASP binding in vitro, have led to the conclusion that PSTPIP can be phosphorylated on tyrosine 367 (19). Furthermore, a model has emerged suggesting that the phosphorylation of PSTPIP on tyrosine 367 is done via a sequential mechanism involving the prior recruitment of a SH2 domaincontaining tyrosine kinase to tyrosine 344 of PSTPIP (19). In this article, we indeed provide some evidence that SH2 domaincontaining tyrosine kinases can interact with PSTPIP in vitro. However, our in vivo co-immunoprecipitation assays suggest either that phosphorylation of PSTPIP on tyrosine 367 is not negatively affecting the interaction with WASP or that tyrosine 367 is simply not phosphorylated in vivo. To resolve this issue, we performed tryptic phosphopeptide mapping of PSTPIP. HEK293-T cells were transfected with PSTPIP or its Y344F and Y367F mutants together with either c-Abl or v-Src. The cells were labeled in vivo with 32 P and PSTPIP was immunoprecipitated, transferred to PVDF, and digested with trypsin. The peptides were recovered and separated by electrophoresis and TLC. As seen in Fig. 7A, two major peptides of PSTPIP (Y1 and Y2) are phosphorylated by c-Abl. Both of these peptides are not phosphorylated in the PSTPIP Y344F mutant co-expressed with c-Abl (data not shown). Interestingly, the Y1 and Y2 peptide are still being phosphorylated by c-Abl in the PSTPIP Y367F mutant (Fig. 7B). We obtained similar results when v-Src was transfected (Fig. 7, C and D). Y2 (c-Abl map) and Y3 (v-Src map) do not migrate identically on TLC. By deduction, Y1 must include tyrosine 344. Importantly, we conclude that PSTPIP is not phosphorylated on tyrosine 367 in vivo. How-ever, we do not exclude the possibility that other tyrosine residues (Y2 and Y3) in PSTPIP are also phosphorylated. The phosphorylation of these tyrosine residues in PSTPIP would have to be dependent on the prior phosphorylation of tyrosine 344.
PSTPIP Serves as a Scaffold Guiding PTP-PEST for a Specific Dephosphorylation of WASP-WASP has been shown to become tyrosine phosphorylated upon adhesion of platelets to collagen, engagement of the B cell-receptor in B-lymphocyte, and IgE cross-linking in mast cells (45)(46)(47)(48). WASP is phosphorylated on tyrosine 291 mainly by two tyrosine kinases in these cells: Lyn and Btk (47,48). The biological function of this tyrosine phosphorylation of WASP has not yet been elucidated. However, recent data on the inactive versus active structure of WASP have provided interesting insight on a putative function of this tyrosine phosphorylation (49). WASP is known to be in an inactive conformation at the resting state. This dormant state of WASP is mediated by an intramolecular binding involving its GTPase-binding domain and the VCA tail. In such a conformation, WASP is unable to interact with the Arp2/3 complex because of its buried VCA tail. This stands as an important mechanism for the regulation of actin nucleation (28). Upon interaction of the inactive WASP with GTP-bound CDC42 via the GTPase-binding domain, WASP undergoes a dramatic conformational change and the VCA domain is exposed and can now interact with the Arp2/3 complex and promote actin nucleation. Interestingly, when WASP is in its inactive conformation, Tyr 291 is buried in an ␣ helix and not readily available for tyrosine phosphorylation. Upon binding of CDC42 and conformational rearrangements, Tyr 291 is now accessible to tyrosine kinases because of the disruption of the ␣ helix. Moreover, it has been proposed that phosphorylation of Tyr 291 could destabilize the inactive conformation of WASP and thus help to keep WASP in an active state (28). In this model, an important aspect of the down-regulation of WASP activity would be the dephosphorylation of Tyr 291 in order for the protein to fold into an inactive conformation in which the VCA is masked in the intramolecular interaction.
We tested if PTP-PEST can dephosphorylate WASP on Tyr 291 . PTP-PEST is unable to interact directly with WASP. Since PSTPIP interacts strongly with both PTP-PEST and WASP, we investigated if it could act as a scaffold between these two molecules. As seen if Fig. 8A, WASP was detected in a PTP-PEST immunoprecipitation. No WASP was detected when the preimmune serum was used. The bridging molecule PSTPIP was also present in the complex as expected (middle panel). The precipitation of PTP-PEST was verified by Western blotting (bottom panel). Flag-WASP becomes tyrosine phosphorylated when the Lyn tyrosine kinase is co-expressed as observed in Fig. 8 (lane 1) and in a previous report (47). We found that PTP-PEST can partially dephosphorylate WASP when PSTPIP is co-transfected (lane 2). However, if a PSTPIP mutant (W232A) unable to interact with PTP-PEST is transfected, WASP is significantly less dephosphorylated by PTP-PEST (lane 3). Furthermore, if a PTP-PEST C231S is transfected, WASP becomes slightly more tyrosine phosphorylated suggesting that the PTP-PEST mutant can interact with and protect WASP in a tyrosine-phosphorylated state. The immunoprecipitation of WASP was verified by reblotting with the anti-Flag antibody (middle panel). Furthermore, the expression of PST-PIP and PTP-PEST proteins was ascertained by Western blotting the cell lysates with the appropriate antibodies. Together, these data suggest that WASP is a substrate of PTP-PEST and that the specificity of this biological event is mediated by the scaffolding action of the adapter protein PSTPIP. This result could have important implications in the regulation of WASP-  D). The cells were labeled with [ 32 P]orthophosphate and PSTPIP was immunoprecipitated from each condition, purified by SDS-PAGE, and subjected to a tryptic digest. The peptides were separated in two dimensions by electrophoresis followed by thin layer chromatography. The phosphopeptides were detected by autoradiography. As observed previously, the Y344F mutant was not phosphorylated and was thus not subjected to phosphopeptide mapping. As seen in A, two phosphorylated peptides (Y1 and Y2) were detected when PSTPIP is phosphorylated by c-Abl. None of these peptides were affected by the mutation Y367F as observed in panel B suggesting that c-Abl does not phosphorylate PSTPIP on tyrosine 367. Similar results were obtained when v-Src was used as a kinase (Y1 and Y3, panels C and D).
like proteins activity toward the Arp2/3 complex and actin polymerization. DISCUSSION We previously noted the hyperphosphorylation of PSTPIP in fibroblasts lacking PTP-PEST suggesting that it might be a substrate for this enzyme (13). In this article, we provide evidence that PTP-PEST and PSTPIP physically interact and that this binding is critical for a specific dephosphorylation of PST-PIP on Tyr 344 . Furthermore, we show that PSTPIP is phosphorylated downstream of the activated PDGF and EGF receptors. This phosphorylation of PSTPIP is most likely mediated by c-Abl since we show that the Src kinases are not involved in this process. We performed phosphopetides maps to clearly demonstrate that PSTPIP is not phosphorylated on Tyr 367 . This is an important finding since it suggests that the association between PSTPIP and WASP would not be regulated by the phosphorylation of this residue in vivo. Interestingly, we show that PSTPIP can act as a linker between PTP-PEST and WASP leading to a specific dephosphorylation of WASP.
The analysis of the expression of the mRNA of PSTPIP by Northern blotting suggests an enrichment in lymphoid organs (18). However, PTP-PEST is known to be ubiquitously expressed (4,6,40). We studied the distribution of PSTPIP at the protein level and we found that it is widely expressed in several murine tissues (spleen, thymus, brain, lung, and muscle) and hematopoietic cells (pro-B cells, macrophages and less abundantly in T-cells). We also detected PSTPIP in NIH 3T3 fibroblasts. The expression of another PTP known to target PSTPIP, PTP-HSCF, is restricted to hematopoietic stem cells (18,50). These protein expression patterns clearly suggest that a more widely expressed PTP must replace PTP-HSCF in other cell types to regulate the phosphorylation of PSTPIP, we propose that PTP-PEST is this candidate. Our data suggests that PTP-PEST and PSTPIP patterns of expression overlap significantly and that the study of their interactions is of biological significance.
Endogenous PSTPIP and PTP-PEST could be co-precipitated from NIH 3T3 fibroblasts and 70Z/3 pro-B cells suggesting that these proteins are included in the same complexes in vivo. We further demonstrated that the carboxyl-terminal homology (CTH) domain of PTP-PEST is required to mediate the association with PSTPIP both in a co-precipitation experiment and in an in vitro binding assay. We conclude that the CTH domain of PTP-PEST is necessary for the binding to PSTPIP. Interestingly, PSTPIP was originally cloned as a PTP-HSCF interacting protein and the CTH domain of PTP-HSCF was shown to be responsible for the interaction (18). Since the CTH domain of PTP-PEST is 75% identical to the one of PTP-HSCF, our results are in agreement with an observation by Spencer et al. (18) that a PTP-PEST peptide containing the CTH domain can disrupt the association between PSTPIP and PTP-HSCF. Conversely, our results demonstrate that the coiled-coil domain of PSTPIP is involved in mediating the interaction with PTP-PEST both in vivo and in vitro. A mutation of a tryptophan residue in the coiled-coil domain of PSTPIP (W232A) disrupts the association between PSTPIP and PTP-PEST. This same mutant of PSTPIP is also impaired in its ability to interact with PTP-HSCF, but it still localizes properly with the actin cytoskeleton suggesting no major disruption of the protein folding (43). Hence, the interactions between PTP-PEST and PSTPIP are conserved at the molecular level when compared with the association of PSTPIP with PTP-HSCF. It is expected that the other PEST-like PTPases PEP and BDP will also bind PSTPIP in an identical manner.
To support our data on the physical association between PTP-PEST and PSTPIP, we found that both proteins co-localize in COS-1 cells by immunofluorescence analysis. This co-localization was significantly reduced when the CTH deletion mutant of PTP-PEST was studied. A surprising observation from those experiments is that the cellular localization of PSTPIP is dramatically affected whether it is bound or not to PTP-PEST. We found that PSTPIP is present at the cortical actin when co-expressed with PTP-PEST ⌬CTH. However, most of the PSTPIP is found in the cytoplasm and the perinuclear region when co-transfected with the WT PTP-PEST. This effect of PTP-PEST on the PSTPIP localization could be a mechanism that regulates the function of PSTPIP by promoting its turnover at the cortical actin. Interestingly, this result also suggests that in future studies the CTH domain of PTP-PEST could be used as a dominant-negative mutant that would pre- FIG. 8. PSTPIP act as a scaffold guiding PTP-PEST for a specific dephosphorylation of WASP. A, PTP-PEST was immunoprecipitated from RAW macrophage lysate. The co-precipitation of WASP and PSTPIP was analyzed by Western blotting. The preimmune antiserum was used as a control. B, Lyn has been previously described as a kinase capable of phosphorylating WASP on tyrosine 291. In conditions in which WASP, Lyn, and PSTPIP were co-expressed, WASP was detected in a tyrosine-phosphorylated state. When PTP-PEST was transfected, a significant decrease in WASP tyrosine phosphorylation was noted but this effect of PTP-PEST could be counteracted by the transfection of the PSTPIP W232A mutant, which is unable to interact with PTP-PEST. WASP phosphorylation was slightly increased in the presence of the C231S mutant of PTP-PEST. vent the adequate localization of PSTPIP.
During this study, we observed that PSTPIP is not significantly tyrosine phosphorylated in proliferating NIH 3T3 fibroblasts. We found that PDGF and EGF receptor activation can result in an increase in PSTPIP tyrosine phosphorylation. Both PDGF and EGF receptors stimulation lead to the activation of several downstream kinases. In both pathways, the Src family of kinases, c-Abl and FAK are known to play important roles (51)(52)(53). We found that c-Abl, c-Src, and Fyn are competent to phosphorylate PSTPIP in a transient transfection assay. However, the PP2 inhibitor did not block the EGF-induced phosphorylation of PSTPIP strongly suggesting that the Src kinases are not involved in the phosphorylation of PSTPIP downstream of the EGF receptor. During the completion of this work, Cong et al. (15) found that PSTPIP becomes tyrosine phosphorylated in mouse fibroblasts following PDGF treatment. This phosphorylation was mainly under the control of c-Abl since PSTPIP failed to become tyrosine phosphorylated in Abl Ϫ/Ϫ fibroblasts (15). To support the hypothesis that c-Abl can directly phosphorylate PSTPIP, it was shown that the SH3 domain of PST-PIP directly interacts with two proline-rich sequences of c-Abl and that this interaction increases the specificity of the kinase reaction (15). From these data, c-Abl is probably the critical tyrosine kinase downstream of the PDGF and EGF receptors involved in PSTPIP tyrosine phosphorylation.
It has been reported that two major sites of tyrosine phosphorylation exist on PSTPIP: Tyr 344 and Tyr 367 (19). Indeed, it was demonstrated that PSTPIP is phosphorylated on tyrosine 344 in cells treated with the PTP inhibitor pervanadate or cells transfected with v-Src since a point mutation of this amino acid to a phenylalanine completely prevented PSTPIP tyrosine phosphorylation (19). However, phosphorylation of PSTPIP on tyrosine 367 was proposed based on the observation that when Y367F mutant was subjected to SDS-PAGE, a slower migrating band of PSTPIP was not detected in Western blotting with an anti-phosphotyrosine antibody (19). We did not detect any differences between the tyrosine phosphorylation levels of PST-PIP WT and Y367F when these proteins were co-expressed with tyrosine kinases. To clarify these findings, we performed an extensive analysis of PSTPIP tyrosine phosphorylation. We found that PSTPIP is not phosphorylated on tyrosine 367 (see Fig. 7). We conclude that the major site of phosphorylation in PSTPIP is Tyr 344 . However, we observed two phosphopeptides for PSTPIP in our maps. We do not rule out the possibility that other tyrosine residues could be phosphorylated in PSTPIP. However, these other phosphorylation sites would have to be dependent on the prior phosphorylation of Tyr 344 . This hypothesis is currently being tested. 2 Tyrosine 344 of PSTPIP is found in a YASI motif. We tested several SH2 domains for their ability to bind phosphorylated PSTPIP and we found that the SH2 domains of c-Abl, and of the Src kinases c-Src and Fyn are capable of such a function in vitro. We also show that mutation of tyrosine 367 of PSTPIP does not affect the binding of the SH2 domain of the tyrosine kinases to PSTPIP, but that mutation of Tyr 344 completely prevented the association with these SH2 domains. These results suggest that Tyr 344 is the binding site for these SH2 domains. The analysis of the Tyr 344 motif, YASI, suggests that it does share some similarities with known c-Src and Fyn SH2 domain-binding motifs. For example, the SH2 domains of c-Src and Fyn can strongly interact with a YAEI motif found in FAK (54,55) suggesting that a YAxI motif could be sufficient for mediating the association of these SH2 with PSTPIP. The biological functions of the complexes formed between PSTPIP and SH2 domain-containing tyrosine kinases may be to transmit the signals from activated EGF and PDGF receptor. However, we failed to detect Src in PSTPIP immunoprecipitate. Thus, further studies are required to determine if PSTPIP can interact with SH2 domain-containing proteins in vivo.
Having established that PSTPIP becomes tyrosine phosphorylated in vivo following growth factor stimulation, we tested if PSTPIP can be a substrate for the catalytic activity of PTP-PEST. We found that Tyr 344 of PSTPIP is a specific site targeted by PTP-PEST. Furthermore, we determined that the binding of PSTPIP to the PTP-PEST CTH domain is essential for an efficient dephosphorylation of PSTPIP. This result indicates that the catalytic domain of PTP-PEST possesses no intrinsic specificity toward phosphorylated PSTPIP, and that PTP-PEST requires another interaction for this enzyme-substrate recognition. This is in contrast to PTP-PEST recognition of tyrosine-phosphorylated p130 Cas in which the PTP domain of PTP-PEST demonstrates a marked specificity and is sufficient to recognize p130 Cas (3,5).
The interaction between PSTPIP and WASP was proposed to be regulated by the phosphorylation of tyrosine 367 located within the SH3 domain of PSTPIP (56). However, this observation could not be reproduced under in vivo situations and we also failed to detect phosphorylation of Tyr 367 in vivo. In this respect, the stability of the PSTPIP⅐WASP complex in vivo was not affected by the co-transfection of wild-type tyrosine kinases such as c-Abl and Fyn. We conclude that it is unlikely that the association of PSTPIP and WASP will be regulated by tyrosine phosphorylation in vivo. Nonetheless, we found that the PST-PIP-WASP association is completely abolished in v-Src expressing cells and that this event is independent of Tyr 367 phosphorylation. The mechanism by which v-Src mediate this effect on the PSTPIP⅐WASP complex is not understood at the moment.
A major interest from the present work is the potential modulation of WASP tyrosine phosphorylation by PTP-PEST. In cells, WASP and N-WASP interaction with the Arp2/3 is tightly regulated by an autoinhibition mechanism of the WASP proteins (28,29). In fact, when native WASP is purified to homogeneity from bovine thymus, it behaves poorly to activate actin nucleation via the Arp2/3 complex (31). Several independent studies have shown that WASP proteins are maintained inactive by intramolecular interactions involving the WA domain and the GTPase-binding domain of these proteins. In this autoinhibited state, the Arp2/3-binding region is hidden and WASP proteins are not capable of promoting actin nucleation. The best understood mechanism for the activation of WASP proteins involves the simultaneous interaction of the proteins with GTP-loaded Cdc42 and the phospholipid PIP 2 (33). These interactions lead to the disruption of the intramolecular interaction and to the exposition of the Arp2/3-binding site of the WASP proteins. Several other mechanisms of activation of these proteins have also been reported involving SH3 domaincontaining proteins and other interacting protein such as WIP but are not understood in great details (34 -36, 57). The possibility of expressing and purifying full-length WASP and N-WASP or their separate domains from bacteria, and using these proteins for in vitro actin nucleation assays in the presence of the purified Arp2/3 complex have been extremely useful to study the activation mechanisms of the WASP proteins. However, is this intramolecular interaction the only way of regulating WASP proteins? Another mechanism of regulation that has not been studied carefully is the tyrosine phosphorylation of WASP proteins. WASP is phosphorylated on tyrosine 291 following a variety of stimuli such as interaction of platelets with collagen, the engagement of the B-cell receptor in B-lymphocytes and the IgE receptor in mast cells (45)(46)(47)(48). Two tyrosine kinases, Btk and Lyn, are reported to be able of phosphorylating WASP. The phosphorylation of N-WASP has not yet been studied. The biological consequences of the tyrosine phosphorylation of WASP remain completely unknown. However, a recent structural analysis of WASP has provided an interesting hypothesis regarding WASP phosphorylation. In this report, the three-dimensional structure of WASP was elucidated by NMR spectroscopy and provided the first structural evidence of the WASP autoinhibited and activated conformations. Interestingly, tyrosine 291 resides in the ␣ helix 3 in the inhibited conformation of WASP. According to the structural data, this tyrosine residue is buried inside the helix and is not likely to be accessible to tyrosine kinases. However, when WASP is activated, tyrosine 291 is available for phosphorylation by tyrosine kinases. In agreement with this model, the co-expression of activated Cdc42 with WASP facilitates WASP tyrosine phosphorylation by Lyn (47). Kim et al. (49) proposed that the phosphorylation of tyrosine 291 could stabilize WASP in the active conformation by preventing the refolding of helix 3 which is present in the inhibited conformation (49). This could be an important mechanism to prolong the duration of WASP signaling in vivo. It will be important to investigate if recombinant WASP fragments phosphorylated on tyrosine 291 behave differently in in vitro actin nucleation assay.
We found that PSTPIP can form a trimolecular complex with PTP-PEST and WASP. Since PSTPIP can interact both with WASP and PTP-PEST, we tested if WASP is a substrate of PTP-PEST. We found that PTP-PEST can dephosphorylate WASP. Furthermore, we uncovered that the scaffolding role of PSTPIP is the critical determinant for the specificity of WASP dephosphorylation by PTP-PEST. Indeed, when a mutant of PSTPIP that has lost its ability to interact with PTP-PEST was transfected, WASP was significantly less dephosphorylated by PTP-PEST. A similar mode of action of PSTPIP was recently shown for the PEST-type PTP-mediated dephosphorylation of the tyrosine kinase c-Abl (15). At the moment, it is difficult to evaluate what is the impact of PTP-PEST action on WASP since the function of the phosphorylation of tyrosine 291 in WASP is still unknown. Elucidating the consequence of the tyrosine phosphorylation of WASP should certainly be a priority. One possibility is that following dephosphorylation by a PTP, WASP is more likely to adopt an inhibited conformation. We provide the first evidence of a PTP, PTP-PEST, in regulating WASP tyrosine phosphorylation. We also observed that PSTPIP tightly interacts with N-WASP both in vivo and in vitro. 3 It will be important to determine if N-WASP also becomes tyrosine phosphorylated in vivo and if it is a target of PTP-PEST. Furthermore, it remains to be investigated if the binding of the PSTPIP SH3 domain to N-WASP or WASP proteins can modulate their actin nucleation activity.