Physical and Functional Interactions between Receptor-like Protein-tyrosine Phosphatase α and p59 fyn *

We have examined the in vivo activity of receptor-like protein-tyrosine phosphatase α (PTPα) toward p59 fyn , a widely expressed Src family kinase. In a coexpression system, PTPα effected a dose-dependent tyrosine dephosphorylation and activation of p59 fyn , where maximal dephosphorylation correlated with a 5-fold increase in kinase activity. PTPα expression resulted in increased accessibility of the p59 fyn SH2 domain, consistent with a PTPα-mediated dephosphorylation of the regulatory C-terminal tyrosine residue of p59 fyn . No p59 fyn dephosphorylation was observed with an enzymatically inactive mutant form of PTPα or with another receptor-like PTP, CD45. Many enzyme-linked receptors are complexed with their substrates, and we examined whether PTPα and p59 fyn underwent association. Reciprocal immunoprecipitations and assays detected p59 fyn and an appropriate kinase activity in PTPα immunoprecipitates and PTPα and PTP activity in p59 fyn immunoprecipitates. No association between CD45 and p59 fyn was detected in similar experiments. The PTPα-mediated activation of p59 fyn is not prerequisite for association since wild-type and inactive mutant PTPα bound equally well to p59 fyn . Endogenous PTPα and p59 fyn were also found in association in mouse brain. Together, these results demonstrate a physical and functional interaction of PTPα and p59 fyn that may be of importance in PTPα-initiated signaling events.

Members of the Src family of tyrosine kinases have been implicated in a variety of physiological and pathophysiological processes. These include mediating mitogenic responses initiated by growth factor receptors, control of cellular architecture through cytoskeletal reorganization, the UV and stress response, mitotic functions, and the induction of tumors (for review, see Ref. 1). While the biological roles of the Src family kinases are not known, it is well established that the activities of these kinases are regulated, in part, by the phosphorylation state of the negative regulatory tyrosine residue corresponding to Tyr-527 of p60 c-src (reviewed in Refs. 2 and 3). Phosphorylation of this residue by Csk or Csk-like kinases represses catalytic activity (4). Preventing phosphorylation of this residue either by association with the polyoma virus middle Tantigen or by mutation to phenylalanine or dephosphorylation of this residue by protein-tyrosine phosphatases results in increased catalytic and transforming activity (5)(6)(7)(8)(9)(10). The identity of such phosphatases is by and large unknown. As mentioned above, the hematopoietic cell protein-tyrosine phosphatase (PTP) 1 CD45 regulates the phosphorylation state and activity of p56 lck and p59 fyn in T cells (11)(12)(13)(14)(15). Presumably, there are other PTPs that regulate the Src family kinases in cells lacking CD45. One possible candidate is PTP␣, a receptor-type PTP.
PTP␣ is a widely expressed protein that differs from most other receptor-like PTPs in having a very short extracellular domain with no adhesion motifs (16 -19). Overexpression of PTP␣ leads to cell transformation and to neuronal differentiation in rat embryo fibroblasts and in P19 carcinoma cells, respectively (20,21). This is similar to the actions of overexpressed epidermal growth factor receptor in A431 and PC12 cells (22,23), suggesting that PTP␣ may normally play a role in stimulating cell proliferation. The intracellular mediators of PTP␣ signaling are not known. The tyrosine kinase pp60 c-src is a candidate PTP␣ substrate since PTP␣ overexpression in rat embryo fibroblasts and P19 cells results in pp60 c-src dephosphorylation and activation (20,21). PTP␣ may also exert some of its cellular effects through its ability to bind the adaptor protein Grb2 (24 -27). Downstream components in a PTP␣ signaling pathway may include mitogen-activated protein kinase and the transcription factor c-Jun, both of which are activated in PTP␣-overexpressing rat embryo fibroblast cells (28). Whether PTP␣ mediates the dephosphorylation of other cellular proteins besides pp60 c-src is unknown.
The similar structure and mode of regulation of Src family kinases suggest that other members of this family may be PTP␣ substrates. The identification of PTP␣ substrates is an important step in elucidating the biological role of PTP␣. In this study, we have investigated the action of PTP␣ toward p59 fyn , prompted by a combination of reasons. First, besides pp60 c-src , only the Src family kinases p59 fyn and p62 yes share a broad expression pattern with PTP␣ (1). In addition, PTP␣ is highly expressed in brain (18,29), and PTP␣, pp60 c-src , and p59 fyn are implicated in or associated with certain neuronal cell functions including neuronal differentiation (PTP␣ (21,30) and pp60 c-src (31,32)), axonal growth (pp60 c-src (33) and p59 fyn (34)), myelination (p59 fyn (35)), and spatial learning and memory (p59 fyn (36)). Second, together with pp60 c-src , p59 fyn is well defined in terms of its cellular actions. While studies in mutant mice show that Src and Fyn kinases have a high degree of functional redundancy (37), nevertheless, they also have specific and distinct functions (for example, in cytoskeletal organization (38) and adhesion molecule-directed axonal growth (33,34)). It is thus important to define which specific intermediate signaling molecules may mediate a spectrum of PTP␣directed cellular events.

EXPERIMENTAL PROCEDURES
Expression Plasmids-Numbering of the PTP␣ amino acid sequence is according to Krueger et al. (17). The expression vector pXJ41-PTP␣neo, encoding full-length PTP␣, has been described (20). The mutagenesis altering the essential cysteine residues to serine residues in the active sites of PTP␣ (C414S/C704S) has been described (39), and a fragment of this cDNA was used to replace the corresponding piece of wild-type PTP␣ cDNA to produce pXJ41-PTP␣(C414S/C704S)-neo. Vectors expressing VSVG-tagged versions of PTP␣ were constructed as follows. Primers (with PacI sites added) corresponding to the amino acid sequences RVGIHL and MNRLGK found at either end of a 29amino acid C-terminal fragment of VSVG (40) were used in a polymerase chain reaction with VSVG cDNA (a gift from Dr. S. H. Wong) as template. The primer sequences were 5Ј-GCGGTTAATTAACCGAGTT-GGTATTTATCTT-3Ј (forward) and 5Ј-GCGGTTAATTAACTTTCCAA-GTCGGTTCAT-3Ј (reverse). The amplified fragment was inserted into a unique PacI site in the PTP␣ cDNAs of pXJ41-neo, permitting the expression of PTP␣ proteins with a VSVG tag at amino acid 16 in the extracellular region. Plasmid containing CD45 cDNA (pAW-HCLA) was a gift of Dr. G. Koretzky. The CD45 cDNA insert was removed with HindIII and subcloned into the HindIII site of pXJ41-Hy (pXJ41 containing the gene conferring hygromycin resistance). fyn cDNA was isolated from a human fetal brain cDNA library in gt11 (CLONTECH, Palo Alto, CA) and subcloned into the EcoRI site of pXJ41-neo. Murine neuronal c-src cDNA (NcoI fragment in pGEM5Z(ϩ); Promega) was a gift of Dr. P. Bello. The c-src cDNA insert was removed with SphI and SacI and blunt end-ligated into the EcoRI site of pXJ41-neo.
Cell Culture and Transient Transfections-COS-1 cells were obtained from American Type Culture Collection (Rockville, MD). Cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and penicillin/streptomycin in an atmosphere of 5% CO 2 at 37°C. COS-1 cells at 50 -70% confluency (100-mm dishes) were transfected with plasmid DNA by liposome-mediated transfection with 30 l (1 mg/ml) of Lipofectin™ or LipofectAMINE™ reagent (Life Technologies, Inc.) for 5-6 h as described by the manufacturer and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum for an additional 18 -40 h prior to harvesting. The empty expression plasmid pXJ41-neo was used to normalize the amount of DNA in each transfection.
Western Blots and Immunoprecipitations from Transfected Cells-In experiments that did not involve the association of PTP␣ (or CD45) and p59 fyn , cell extracts were prepared by lysing cells either in buffer A (50 mM Tris-Cl (pH 7.2), 150 mM NaCl, 0.2 mM Na 3 VO 4 , 1% Triton X-100, 10 g/ml aprotinin, and 0.1 mM phenylmethylsulfonyl fluoride) (see Figs. 2 and 4C) or in modified radioimmune precipitation assay buffer (10 mM sodium phosphate (pH 7.0), 150 mM NaCl, 1 mM EDTA, 50 mM NaF, 0.1 mM Na 3 VO 4 , 1% Nonidet P-40, 0.1% SDS, 1% sodium deoxycholate, 10 g/ml aprotinin, and 0.1 mM phenylmethylsulfonyl fluoride) (see Figs. 1 and 3) for 60 min at 4°C. Cytosol and Triton X-100-solubilized membrane fractions of cells were obtained as described (20), essentially involving initial lysis of cells by sonication in buffer A without Triton X-100. In experiments involving co-immunoprecipitation of PTP␣ (or CD45) and p59 fyn , cell lysates were prepared in 10 mM Tris-Cl (pH 7.2), 150 mM NaCl, 1 mM EDTA, 1% Brij 96, 10 g/ml aprotinin, and 0.1 mM phenylmethylsulfonyl fluoride (see Figs. 5-7). Lysates were clarified by centrifugation, and protein content was determined by Bradford analysis (41). Protein extracts were separated by SDS-polyacrylamide gel electrophoresis on a 7 or 9% gel and electrophoretically transferred to polyvinylidene difluoride membranes. Membranes were immunoblotted with anti-PTP␣ antiserum 3897 (raised against a glutathione S-transferase/PTP␣-D2 fusion protein (39) containing the second catalytic domain and C-terminal tail region of PTP␣) followed by goat anti-rabbit IgG conjugated to peroxidase (Sigma) or with anti-VSVG (Sigma), anti-p59 fyn (Transduction Laboratories), anti-pp60 c-src (Oncogene Science Inc.), or anti-CD45 9.4 (Dr. G. Koretzky) monoclonal antibodies followed by goat anti-mouse IgG conjugated to peroxidase (Sigma) or peroxidaseconjugated anti-phosphotyrosine antibody (Transduction Laboratories). Immunoblots were developed using the ECL system (Amersham Pharmacia Biotech). Immunoprecipitations of cell lysates were carried out using 250 -600 g of protein. For immunoprecipitation of p59 fyn , either anti-Fyn antiserum (a gift of C. Rudd; see Fig. 1) or anti-Fyn polyclonal antibody (FYN3-G, Santa Cruz Biotechnology, Inc.) was added to the cell lysates and incubated for 60 -120 min at 4°C. For pp60 c-src and VSVG-PTP␣ immunoprecipitations, anti-pp60 c-src or anti-VSVG monoclonal antibodies were added to the cell lysates and incubated for 60 min at 4°C, followed by incubation with rabbit anti-mouse IgG (Dako Corp.) for another 60 min at 4°C. Protein A cell suspension (Sigma) was then added and mixed at 4°C for 1 h. The immunoprecipitates were washed three times each in the respective cell lysis buffer (see above) and once in either 2ϫ kinase assay buffer or phosphatase assay buffer (see below). Immunoblot analysis of the immunoprecipitated proteins was as described above.
Immunoprecipitations from Mouse Brain Lysate-Whole brain from an adult BALB/c mouse was homogenized in a hand-held Wheaton homogenizer in 6 -8 ml of 10 mM Tris-Cl (pH 7.2), 150 mM NaCl, 1 mM EDTA, 1% Brij 96, 10 g/ml aprotinin, and 0.1 mM phenylmethylsulfonyl fluoride. After incubation at 4°C for 90 min, the lysate was clarified by centrifugation, and the protein content was determined. For immunoprecipitations, 1 mg of lysate was diluted a further 4-fold in homogenization buffer and then incubated with anti-PTP␣ (3680, raised against a peptide comprising the C-terminal 18 amino acids of PTP␣), anti-p59 fyn (FYN3, Santa Cruz Biotechnology, Inc.), or anti-Csk (C-20, Santa Cruz Biotechnology, Inc.) polyclonal antibodies. In some experiments, anti-PTP␣ antibody 3680 was blocked before immunoprecipitation by preincubation with recombinant purified PTP␣-D2 polypeptide (amino acids 485-774) for 45 min at 4°C. The brain lysate was incubated with the above antibodies for 16 h at 4°C, and then Protein G PLUS/Protein A-agarose (Calbiochem) was added, and incubation was continued for 60 min at 4°C. The immunoprecipitates were washed three times each in lysis buffer containing Brij 96 and once in lysis buffer without detergent, resolved by 8.5% SDS-polyacrylamide gel electrophoresis, and transferred to polyvinylidene difluoride membranes. Membranes were immunoblotted with anti-PTP␣ antibody 3897 or 3680 followed by goat anti-rabbit IgG conjugated to peroxidase. Immunoblots were developed using the SuperSignal chemiluminescence system (Pierce).
Accessibility of the p59 fyn SH2 Domain-Synthetic peptides with the sequence TSTEPQYQPGENL, representing the sequence surrounding Tyr-527 of pp60 c-src , were made using either phosphotyrosine or tyrosine at the appropriate step and were purified by high pressure liquid chromatography (Biotechnology Center, National University of Singapore). The phosphopeptide or peptide was covalently coupled to CNBr-activated Sepharose-4B (Amersham Pharmacia Inc.) and added to 300 g of whole cell lysates (prepared with radioimmune precipitation assay buffer as described above) of COS-1 cells transfected with p59 fyn cDNA alone or in combination with PTP␣ cDNA. After incubation for 2 h at 4°C, the Sepharose beads were washed twice each with radioimmune precipitation assay buffer in the presence and absence of SDS/sodium deoxycholate, respectively, and resolved by electrophoresis on a 10% SDS-polyacrylamide gel. Immunoblot analysis was as described above.
Phosphatase Assays-In experiments where PTP activity was measured, Na 3 VO 4 was omitted from the cell lysis buffer. PTP activity was measured at 30°C in reactions containing 50 mM Mes (pH 6.0), 0.5 mg/ml bovine serum albumin, 0.5 mM dithiothreitol, and 2.5-5 M phosphotyrosyl-RR-Src peptide (RRLIEDAEY(P)AARG, corresponding to the sequence encompassing Tyr-416 of pp60 c-src and phosphorylated as described (39)). The specific activity of the substrate ranged between 3000 and 4000 cpm/pmol of RR-Src. The reaction was carried out for 3 min unless otherwise indicated.

PTP␣ Effects
Dephosphorylation of pp60 c-src and p59 fyn -The ability of PTP␣ to dephosphorylate pp60 c-src and p59 fyn was assessed in COS-1 cells cotransfected with both kinases. Antiphosphotyrosine probing of p59 fyn and pp60 c-src immunoprecipitates demonstrated that PTP␣ expression resulted in tyrosine dephosphorylation of both kinases (Fig. 1, A and B). Blotting of these cell lysates with anti-cst.1, an antibody that recognizes both p59 fyn and pp60 c-src equally well (42), showed that similar amounts of these kinases were expressed with PTP␣ (data not shown). As p59 fyn represents a novel potential substrate for PTP␣, we characterized the catalytic action of PTP␣ toward p59 fyn in more detail. The extent of p59 fyn dephosphorylation increased as increasing amounts of PTP␣ cDNA were transfected, reaching a plateau of 80 -90% tyrosine dephosphorylation ( Fig. 2A). Dephosphorylation was completely dependent on the catalytic activity of PTP␣ since a PTP␣ mutant, PTP␣(C414S/C704S), in which the essential cysteine residues in the active sites of both catalytic domains of PTP␣ were mutated to serine residues, was unable to effect p59 fyn dephosphorylation ( Fig. 2A).
Coexpression of PTP␣ and p59 fyn Results in p59 fyn Activation-Besides tyrosine dephosphorylation of p59 fyn , the coexpression of PTP␣ resulted in kinase activation of p59 fyn . As shown in Fig. 2B, when p59 fyn immunoprecipitates from cells coexpressing PTP␣ were used in an immunocomplex kinase assay, increasing p59 fyn autophosphorylation was observed with increasing expression of PTP␣. A maximum 5-fold increase in kinase activity was obtained (Fig. 2B, lanes 5 and 6) at the same PTP␣/p59 fyn cDNA ratio observed to give maximal p59 fyn dephosphorylation ( Fig. 2A). No increase in the kinase activity of p59 fyn was measured when catalytically inactive PTP␣ was expressed with p59 fyn (data not shown).
p59 fyn Dephosphorylation and Activation Are Accompanied by Increased SH2 Domain Accessibility-Dephosphorylation of the C-terminal tyrosine residue of Src family kinases is linked to kinase activation (5-9) and correlates with increased c-Src SH2 domain and Lck SH2 domain accessibility during mitosis and T cell activation, respectively (43,44). These observations support a model of kinase activation where disruption of an intramolecular association between the C-terminal phosphotyrosyl peptide and the SH2 domain of the kinase results in catalytic activation as well as novel interactions of the SH2 domain with other phosphotyrosyl proteins (2,3). The effect of PTP␣ on p59 fyn SH2 domain accessibility was examined by determining the ability of p59 fyn to bind to a synthetic phosphopeptide representing the C-terminal Tyr-527 peptide of pp60 c-src . This peptide is identical to the C-terminal sequence surrounding Tyr-531 of p59 fyn except for the replacement of alanine in position 2 with serine (45). As shown in Fig. 3, PTP␣-induced dephosphorylation of p59 fyn (Fig. 3C) correlated with a 3-fold increase in the amount of Fyn protein precipitated from cell lysates with the Src Tyr-527 phosphopeptide coupled to Sepharose beads (Fig. 3D, bottom panel). A control precipitation using an unphosphorylated Src Tyr-527 peptide-Sepharose conjugate contained barely detectable but equivalent amounts of Fyn protein from p59 fyn -and PTP␣/p59 fyn -expressing cells (Fig. 3D, top panel). While the site(s) of p59 fyn dephosphorylation remain to be mapped, the increased p59 fyn catalytic activity and SH2 availability for binding are consistent with a PTP␣-mediated dephosphorylation of the C-terminal Tyr-531 of p59 fyn .
Specificity of PTP␣ in Effecting p59 fyn Dephosphorylation and Activation-To examine whether the PTP␣-mediated p59 fyn dephosphorylation was a specific effect of PTP␣ or merely reflected a nonspecific increase in PTP activity at the cell membrane, the tyrosine phosphorylation state of p59 fyn was analyzed upon coexpression with another receptor-like COS-1 cells were transfected with empty plasmid (mock), 3 g each of p59 fyn and pp60 c-src cDNAs together, or 3 g each of p59 fyn , pp60 c-src , and PTP␣ cDNAs together, with the total amount of DNA in each transfection equalized by the addition of empty plasmid. Immunoprecipitates of p59 fyn (A) or pp60 c-src (B) from 250 g of the cell lysates were probed with anti-phosphotyrosine (top panels) and anti-p59 fyn (A) or pp60 c-src (B) (bottom panels) antibodies. HC, heavy chain antibody. PTP, CD45. CD45 is a hematopoietic cell-specific molecule required for T and B cell activation (46 -48) and can dephosphorylate the Src family kinases p59 fyn and p56 lck in T cells (11)(12)(13)(14)(15) and p62 lyn in B cells (49). Fractionation of transfected COS cells into solubilized membrane and cytosol followed by Western blotting showed that, like PTP␣ (Fig. 4A, middle panel), CD45 is localized to membranes (bottom panel). As expected, a majority of the Fyn protein was also associated with membranes (Fig. 4A, top panel). Elevated membrane PTP activity was measured in PTP␣-or CD45-expressing cells (Fig. 4B), demonstrating that both receptor PTPs are enzymatically active and that comparable levels of phosphatase activity are seen in both cell types. However, immunoprecipitated p59 fyn was tyrosine-dephosphorylated in PTP␣-expressing cells (Fig.  4C, compare lanes 1 and 2), whereas no reduction in the tyrosine phosphate content of p59 fyn from CD45-expressing cells was detected (compare lanes 1 and 3). Even when the amount of CD45 cDNA was increased 4-fold over that used for the experiments shown in Fig. 4, no dephosphorylation of coexpressed p59 fyn was observed (data not shown). This is similar to another report that CD45 cannot effect p56 lck dephosphorylation when expressed in non-lymphoid cells (15). The p59 fyn dephosphorylation observed in the presence of PTP␣, but not CD45, correlated with elevated kinase activity of Fyn in PTP␣expressing cells (as described above) and no alteration in the kinase activity of p59 fyn from CD45-expressing cells (data not shown). Thus, increased membrane PTP activity is not in itself sufficient to effect p59 fyn dephosphorylation and activation, indicating that the PTP␣-mediated dephosphorylation and activation of p59 fyn reflect a specific effect of PTP␣.
Association of PTP␣ and p59 fyn -To further investigate the interaction of PTP␣ and p59 fyn , we examined whether these two proteins underwent any form of association. To enable immunoprecipitation of PTP␣, a 29-amino acid epitope from VSVG was inserted into the PTP␣ extracellular region to create VSVG-PTP␣. Cells expressing p59 fyn alone or in conjunction with VSVG-PTP␣ were lysed under mild detergent conditions, and PTP␣ was immunoprecipitated with anti-VSVG antibodies. Probing of the immunoprecipitates with anti-VSVG antibodies detected two forms of VSVG-PTP␣: a broad diffuse band migrating at ϳ130 kDa, consistent with the size of wild-type glycosylated PTP␣, and a sharper band at ϳ100 kDa, consistent with the size of incompletely glycosylated PTP␣ (50) (Fig.  5A, top panel, lanes 7 and 8). Probing of these samples and whole cell lysates with anti-p59 fyn antibodies revealed about equal amounts of Fyn protein in cell lysates expressing p59 fyn alone or together with VSVG-PTP␣ (Fig. 5A, bottom panel,  lanes 2 and 4) and a significant amount of p59 fyn in the anti-VSVG immunoprecipitate from the PTP␣/p59 fyn -coexpressing cells (lane 8). The presence of this p59 fyn in the VSVG immunoprecipitate was due to its interaction with VSVG-PTP␣ since only a weak p59 fyn signal (likely due to nonspecific sticking) was detected in anti-VSVG immunoprecipitates prepared from cells expressing p59 fyn in the absence of VSVG-PTP␣ (Fig. 5A,  bottom panel, lane 6). In addition, in vitro assay of kinase activity in the anti-VSVG immunoprecipitates detected enhanced phosphorylation of a protein (Fig. 5B, lane 5) that comigrated with autophosphorylated p59 fyn produced by in vitro kinase assay of an anti-Fyn immunoprecipitate from p59 fyn -expressing cells (lane 1). These and subsequent experiments to examine the association of PTP␣ and p59 fyn were carried out with cell lysates prepared by solubilization with Brij 96, a mild non-ionic detergent. The association of PTP␣ and p59 fyn was also detected in lysates prepared in buffer containing Triton X-100 or in radioimmune precipitation assay buffer, indicating that the interaction is also stable in other non-ionic detergents and in ionic detergents (data not shown). The above results indicate that p59 fyn associates with and can be co-immunoprecipitated with PTP␣.
Reciprocal experiments were carried out to confirm the association of PTP␣ and p59 fyn in which p59 fyn immunoprecipitates from transfected cells were analyzed for the presence of PTP␣. Probing with anti-VSVG antibodies revealed the presence of VSVG-PTP␣ in p59 fyn immunoprecipitates from PTP␣/ p59 fyn -expressing cells (Fig. 6A, lane 6), but not in immunoprecipitates from p59 fyn -or PTP␣-expressing cells (lanes 4 and 5). Although two forms of VSVG-PTP␣ are present in whole cell lysates, likely corresponding to glycosylated and incompletely glycosylated forms of PTP␣ (50), only the larger glycosylated form was found in the p59 fyn immunoprecipitates. It is possible that the less glycosylated PTP␣ is incompletely processed and trapped in the endoplasmic reticulum and/or Golgi apparatus, and its absence in the p59 fyn immunocomplex suggests that PTP␣ and p59 fyn interaction represents a specific cellular protein-protein association that does not merely occur as a postlysis event. When PTP␣ was replaced with CD45 in the transfections, no co-immunoprecipitated CD45 was detected in p59 fyn immunoprecipitates from CD45/p59 fyn -expressing cells (Fig. 6B). The above results correlated with the presence of PTP activity when portions of the p59 fyn immunoprecipitates analyzed in Fig. 6 (A and B) were assayed. Low levels of PTP activity were present in the p59 fyn immunoprecipitates from cells transfected with p59 fyn , PTP␣, or CD45 alone or with p59 fyn and CD45 together. A much higher level of PTP activity was present in the p59 fyn immunoprecipitates from cells coexpressing p59 fyn and PTP␣ (Fig. 6C). The same results were obtained when untagged PTP␣ was used in the experiment (data not shown), indicating that the VSVG tag does not interact with p59 fyn . Thus, despite the higher PTP activity in lysates of CD45-expressing cells compared with PTP␣-expressing cells (Fig. 6C, inset), only activity attributable to PTP␣ co-immunoprecipitated with p59 fyn .
Association of p59 fyn and a Catalytic Mutant of PTP␣-The mechanism and temporal occurrence of PTP␣ and p59 fyn association (pre-or post-dephosphorylation) were examined using a catalytically inactive form of PTP␣. Mutation of the essential cysteine residue in the conserved catalytic domain of PTPs creates an enzymatically inactive PTP (51,52), which has been shown to bind to and "trap" phosphotyrosyl substrates (53)(54)(55). A VSVG-PTP␣ double mutant in which the essential cysteine residues in both the membrane proximal and distal catalytic domains were mutated to serine residues (C414S/C704S) was produced. If association involves the recognition of phosphotyrosine sites in p59 fyn as a pre-dephosphorylation event, then such interaction might be stabilized and enhanced with enzymatically inactive PTP␣. However, if p59 fyn dephosphorylation is prerequisite for PTP␣-p59 fyn association, then the phosphatase-kinase complex will not be formed with the enzymatically inactive mutant. No dephosphorylation of p59 fyn was detected upon coexpression with mutant VSVG-PTP␣(C414S/C704S) (data not shown, but see Fig. 2A). Nevertheless, p59 fyn was detected in immunoprecipitates of VSVG-PTP␣(C414S/C704S) at a level equivalent to that of p59 fyn in immunoprecipitates of catalytically active VSVG-PTP␣ (Fig. 7). Thus, a conventional substrate-trapping technique fails to enhance PTP␣-p59 fyn interaction, suggesting that additional or alternative regions of these proteins are responsible for association. Also, the finding that p59 fyn and PTP␣ can associate independently of PTP␣ activity, and thus in the absence of or prior to dephosphorylation, indicates that PTP␣ is suitably positioned to directly utilize p59 fyn as a substrate.
Association of Endogenous PTP␣ with p59 fyn -The above studies were carried out with COS cells ectopically expressing PTP␣ and p59 fyn . To determine if association occurred between these proteins when present at endogenous levels, we examined whether they could be co-immunoprecipitated from mouse brain lysates. Brain was chosen as the tissue source because PTP␣ is highly expressed in brain (18,29) and because neuronal functions of both PTP␣ (21,30) and p59 fyn (34 -36) have been reported. None of the PTP␣ antibodies available to us immunoprecipitated PTP␣ very efficiently, based on comparisons with the level of PTP␣ detected by Western blotting of brain lysates (data not shown). Some PTP␣ could be precipitated (Fig. 8, lane 2) with a polyclonal antibody (3680) raised against a synthetic peptide corresponding to the C-terminal 18 amino acids of PTP␣, and specific immunoprecipitation of PTP␣ was blocked by preincubation of the antibody with recombinant purified protein (PTP␣-D2) comprising the membrane distal catalytic domain and C terminus of PTP␣ (Fig. 8,  lane 1). PTP␣ was detected in p59 fyn immunoprecipitates (Fig.  8, lane 3), but not in immunoprecipitates of Csk (lane 4), a non-receptor tyrosine kinase structurally similar to p59 fyn . The above immunoprecipitations were probed with anti-PTP␣ antibody (3897) raised against recombinant PTP␣-D2. Stripping and reprobing of lanes 1-4 with the anti-PTP␣ C-terminal antibody (3680, used for PTP␣ immunoprecipitations; see above) gave the same results (data not shown). This association of PTP␣ with p59 fyn was observed in repeated experiments and demonstrates that these proteins are physiologically associated. DISCUSSION We have confirmed that PTP␣ dephosphorylates pp60 c-src in vivo and found that PTP␣ can also dephosphorylate the related Src family kinase p59 fyn . The concomitant increase in p59 fyn kinase activity and increase in accessibility of the p59 fyn SH2 domain are consistent with a PTP␣-mediated dephosphorylation of the C-terminal tyrosine residue thought to be critical in regulating the activity of p59 fyn . In contrast, the expression of the receptor-like PTP CD45 in this system did not result in p59 fyn dephosphorylation or activation, demonstrating a specific action of PTP␣ on p59 fyn . Previously, PTP␣ activity has mainly been assayed for artificial substrates in vitro using immunoprecipitated PTP␣ or bacterially expressed forms of cytosolic PTP␣ (24,39,56,57). This characterization of a cellular assay of PTP␣ activity will be useful in the future assessment of the effects of various structural mutants of PTP␣ on its enzymatic function.
An important question regarding the activation of Src family kinases by PTP␣ is whether these kinases are directly dephosphorylated by PTP␣ and thus represent PTP␣ substrates. Potential PTP␣ substrates might be complexed with the phosphatase, as is the case with many tyrosine kinases and their substrates, or with CD45 and its substrates in T (58 -62) and B (63) cells. Co-immunoprecipitation experiments with transfected cells demonstrated that PTP␣ and p59 fyn were consistently found to be associated with one another. This was not merely an artifact of heterologous expression of these proteins since CD45 and p59 fyn were not found in association under the same experimental conditions. Likewise, what appears to be an incompletely glycosylated (and thus perhaps inappropriately localized) form of PTP␣ did not associate with p59 fyn . In addition, PTP␣ was detected in p59 fyn immunoprecipitates from mouse brain. Although PTP␣-p59 fyn association is suggestive of a direct enzyme-substrate relationship, at present it is unclear whether these physical and functional actions of PTP␣ are linked. If so, two possibilities are that association enables the subsequent dephosphorylation and activation of p59 fyn or that activated p59 fyn directly or indirectly modifies PTP␣ to promote PTP␣-p59 fyn association, perhaps after tyrosine phosphorylation of the phosphatase creates p59 fyn -binding sites. Our evidence supports the former scenario since the association of a catalytically inactive form of PTP␣ with p59 fyn indicates that this physical interaction does not require prior dephosphorylation/activation of p59 fyn . We are generating truncated forms of PTP␣ to define the region(s) involved in p59 fyn binding. Regions of p59 fyn that associate with a variety of other signaling molecules include the SH3 and SH2 domains (for examples, see Refs. 64 -67) and the unique N-terminal region (68,69). A proline-rich sequence similar to the consensus sequence for SH3 binding (70 -72) is found in PTP␣ (RKYPPLP, amino acids 188 -194). As PTP␣ is tyrosine-phosphorylated in the cell (25,26), this could provide sites for SH2 binding. Alternatively, PTP␣-p59 fyn association may occur through other regions or intermediary proteins.
Besides the association with p59 fyn described here, PTP␣ can associate with the adaptor molecule Grb2 (24 -27). The latter complex is formed upon phosphorylation of a tyrosine residue in the tail region of PTP␣ and binding by the SH2 domain of Grb2. A direct or indirect association also occurs between the C-terminal SH3 domain of Grb2 and a non-proline-rich region near the active site of the membrane proximal catalytic domain of PTP␣, but is not observed in the absence of the PTP␣-Grb2(SH2) interaction. One effect of PTP␣-Grb2(SH3) binding is postulated to be the inhibition of the catalytic activity of PTP␣ through the obstruction of substrate binding (26). Tyrosine phosphorylation of PTP␣ can be catalyzed by pp60 c-src (24), and it will be of interest to see if p59 fyn can phosphorylate PTP␣ at the same site to result in Grb2 binding. We have observed enhanced tyrosine phosphorylation of PTP␣ upon coexpression with p59 fyn , 2 although the site(s) of phosphorylation is unknown. If this phosphorylation occurs at the appropriate Cterminal site, PTP␣-catalyzed activation of pp60 c-src and/or p59 fyn would activate kinase-mediated downstream signaling events while also permitting the feedback inhibition of PTP␣ by effecting Grb2 binding. Regardless of whether p59 fyn activation and Grb2 binding are linked, it will be of interest to see whether p59 fyn and Grb2 binding to PTP␣ can occur on the same PTP␣ molecule or are mutually exclusive.
Previous studies have suggested that the cell transformation and retinoic acid-induced neuronal differentiation observed in certain cell types upon PTP␣ expression may be mediated through pp60 c-src (20,21). Here we have provided further evidence that PTP␣ is a physiological regulator of the Src family kinases and, in particular, that p59 fyn is a direct in vivo substrate of PTP␣. As with pp60 c-src , prevention of C-terminal Tyr-531 phosphorylation of p59 fyn (by mutation of this tyrosine to phenylalanine) results in an oncoprotein, which upon overexpression transforms rodent fibroblasts (73). The altered phenotype of PTP␣-expressing cells may be due to the increased kinase activity of p59 fyn . Alternatively, PTP␣-induced transformation and differentiation may be a consequence of the combined or synergistic effect of increased p59 fyn and pp60 c-src catalytic activity. The association of PTP␣ with p59 fyn in brain, in conjunction with the presence of both proteins in neuronal cells such as cerebellar granule cells (29,74,75) and dorsal root ganglia (34,76), suggests that p59 fyn could be a component of as yet unknown PTP␣ signaling pathways of neuronal development.
Dephosphorylation and activation of Src family kinases have been demonstrated or implicated in studies of the signaling pathways of two receptor-like PTPs, PTP␣ and CD45. It is conceivable that these and other non-adhesion receptor-type PTPs share a common mode of signaling, with specificity determined by the respective spatial and temporal patterns of gene expression of the receptor-type PTPs and Src family kinases.