Identification of p130 cas as an in VivoSubstrate of Receptor Protein-tyrosine Phosphatase α*

We have employed a substrate trapping strategy to identify physiological substrates of the receptor protein-tyrosine phosphatase α (RPTPα). Here we report that a substrate-trapping mutant of the RPTPα membrane proximal catalytic domain (D1), RPTPα-D1-C433S, specifically bound to tyrosine-phosphorylated proteins from pervanadate-treated cells. The membrane distal catalytic domain of RPTPα (D2) and mutants thereof did not bind to tyrosine-phosphorylated proteins. The pattern of tyrosine-phosphorylated proteins that bound to RPTPα-D1-C433S varied between cell lines, but a protein of approximately 130 kDa was pulled down from every cell line. This protein was identified as p130 cas . Tyrosine-phosphorylated p130 cas from fibronectin-stimulated NIH3T3 cells bound to RPTPα-D1-C433S as well, suggesting that p130 cas is a physiological substrate of RPTPα. RPTPα dephosphorylated p130 cas in vitro, and RPTPα co-localized with a subpopulation of p130 cas to the plasma membrane. Co-transfection experiments with activated SrcY529F, p130 cas , and RPTPα or inactive, mutant RPTPα indicated that RPTPα dephosphorylated p130 cas in vivo. Tyrosine-phosphorylated epidermal growth factor receptor was not dephosphorylated by RPTPα under these conditions, suggesting that p130 cas is a specific substrate of RPTPα in living cells. In conclusion, our results provide evidence that p130 cas is a physiological substrate of RPTPα in vivo.

Protein-tyrosine kinases (PTKs) 1 and protein-tyrosine phosphatases (PTPs) regulate the level of tyrosine phosphorylation of target proteins in cells, thereby regulating many important eukaryotic cell-signaling pathways. PTPs catalyze the hydrolysis of phosphoryl groups on Tyr residues in proteins. Each member of the PTP family contains one or two conserved PTP domains of approximately 240 amino acids including the signature motif (I/V)HCXAGXXR(S/T)G (1). These PTP domains are not only conserved in sequence but also in structure (2)(3)(4)(5)(6)(7)(8).
The PTP family can be subdivided based on structural differences into receptor-like (RPTP) and cytosolic proteins (9). RPTPs, with CD45 as the founding member (10), consist of an extracellular domain, a single membrane spanning domain, and a cytoplasmic domain. Most RPTPs contain two tandemly repeated PTP domains in their cytoplasmic domain. Interestingly, for all RPTPs with two PTP domains, the majority of the catalytic activity resides within the membrane proximal PTP domain (D1), whereas the membrane distal domain, D2, displays little or no catalytic activity. Inactivation of RPTP␣-D1 or CD45-D1 is sufficient to abolish their biological activities, indicating that PTP activity in D1, but not D2, is essential for the function of RPTPs (11,12). D2s are conserved in sequence and in structure (6,13), 2 but all D2s lack residues essential for catalysis, suggesting an important role for D2s in processes other than catalysis (6,14,15).
To elucidate the function of PTPs in vivo, it is essential to know their natural substrates. To identify substrates of PTPs, "substrate trapping" mutants have been designed. For instance, mutation of the catalytic site Cys to Ala in YopH, a Yersinia PTP, resulted in a mutant that bound substrates but was no longer able to dephosphorylate them, making this an excellent tool for identifying potential substrates (16). Since then, other mutants have been found to bind substrates. For instance, mutation of the highly conserved Asp residue from the "WpD" loop that functions as a general acid to facilitate cleavage of the scissile P-O bond in the substrate generated an efficient substrate-trapping mutant. In fact, for PTP-PEST and PTP1B, these Asp mutants were shown to be even more efficient substrate-trapping mutants than the catalytic site Cys mutants (17,18). By now, substrate-trapping mutants of many non-receptor PTPs have successfully been used to identify substrates (17)(18)(19)(20)(21)(22)(23)(24)(25).
RPTP␣ is a transmembrane PTP with a short heavily glycosylated extracellular domain and two cytoplasmic PTP domains. Several potential physiological substrates of RPTP␣ have been identified. Expression of RPTP␣ was shown to interfere with insulin receptor (IR) signaling (26). Other reports have shown that RPTP␣ interferes with insulin-induced prolactin gene expression and GLUT4 translocation to the membrane (27,28). Whether direct dephosphorylation of the IR by RPTP␣ mediates the effects on the IR-signaling pathway remains to be determined definitively. RPTP␣ associates with the Kv1.2 potassium channel in response to activation of the m1 muscarinic acetylcholine receptor, a G protein-coupled receptor. m1 muscarinic acetylcholine receptor activation regulates a tyrosine kinase that phosphorylates the Kv1.2 potassium channel, thereby suppressing the current generated by this channel. RPTP␣ recruitment to the Kv1.2 channel reverses the tyrosine kinase-induced phosphorylation and suppression of Kv1.2, suggesting that the Kv1.2 potassium channel may be a direct substrate of RPTP␣ in cells (29).
The PTK Src is the most clearly defined target of RPTP␣ (11,30). RPTP␣ dephosphorylates and activates Src in vitro and in vivo (11, 30 -32). Murine Src is phosphorylated on Tyr 529 . The crystal structure of Src and Src family members demonstrated that the SH2 domain of Src binds to phosphorylated Tyr 529 , thereby blocking the catalytic site of the kinase (Ref. 33 and references therein). Recently, a phosphotyrosine displacement mechanism was proposed to underlie RPTP␣-mediated dephosphorylation of Src Tyr(P) 529 (34). The C-terminal RPTP␣ Tyr(P) 789 binds the Src SH2 domain, thereby displacing intramolecular Src SH2-Tyr(P) 529 binding and allowing dephosphorylation and thus activation of Src. This model is consistent with the finding that RPTP␣ associates with Src family members independently of RPTP␣ activity (35,36). RPTP␣ knockout mice show an increase in Src Tyr 529 phosphorylation and a decrease in Src activity, providing strong support for the finding that Src is an in vivo substrate of RPTP␣ (31,32). Similarly, targeted disruption of RPTP␣ showed a decrease in activity of the Src family member Fyn (31,32).
We set out to identify physiological substrates of RPTP␣, using substrate-trapping mutants. Here, we report that a substrate-trapping mutant of RPTP␣ specifically bound to tyrosine-phosphorylated p130 cas from pervanadate-treated and fibronectin-stimulated cells. The interaction with RPTP␣ depended on the tyrosine phosphorylation state of p130 cas . Furthermore, RPTP␣ dephosphorylated p130 cas in vitro and in vivo. Analysis of the subcellular localization of p130 cas and RPTP␣ by (immuno)fluorescence demonstrated that some, but not all, p130 cas co-localized with RPTP␣ at the plasma membrane. Our results demonstrate that tyrosine-phosphorylated p130 cas is a substrate of RPTP␣.

MATERIALS AND METHODS
Cells and Transfections-NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum (Life Technologies, Inc.). HepG2 cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal calf serum. SK-N-MC neuroepithelioma and FaO cells were cultured in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F12 (DF) medium supplemented with 10% fetal calf serum. P19 EC, COS-7, and F9 cells were cultured in DF supplemented with 7.5% fetal calf serum.
Pervanadate treatment of the cells was done for 30 min by addition of 1 mM orthovanadate and 1 mM H 2 O 2 , generating pervanadate, directly into the medium of nearly confluent cells. Fibronectin stimulation was done essentially as described (37). Serum-starved NIH3T3 cells were removed from the dish gently using EDTA, allowed to recover in medium supplemented with 0.5% bovine calf serum for 30 min, and replated onto dishes that had been coated with fibronectin (10 g/ml, overnight, 4°C). The cells were harvested after 30 min. Transient and stable transfection of SK-N-MC and P19 EC cells was done using calcium phosphate precipitation, exactly as described (11).
Recombinant Enzymes-The glutathione S-transferase fusion proteins were purified essentially as described (43). The fusion proteins were not cleaved off the beads but were eluted with 10 mM reduced glutathione in 50 mM Tris, pH 8.0, 10% glycerol, 10 units/ml aprotinin, and 1 M PMSF. The fusion proteins were dialyzed against TBS (50 mM Tris, pH 8.0, 150 mM NaCl) and rebound to glutathione-agarose beads. This elution and rebinding to glutathione-agarose beads greatly reduced contamination with bacterial proteins.
Indirect Immunofluorescence-For immunofluorescence labeling, cells were seeded on glass coverslips, and at the appropriate time the cells were fixed with 2% paraformaldehyde in PBS. After washing with PBS the cells were permeabilized with 0.1% Triton X-100 in PBS for 45 s. After washing the coverslips were blocked in 5% bovine serum albumin in PBS for 1 h. Incubation of the coverslips with anti-neurofilament or anti-p130 cas antibodies and CY3-conjugated secondary antibodies was done exactly as described previously (38).

RPTP␣-D1-C433S
Specifically Bound p130 cas -In this study we set out to identify substrates of RPTP␣. Specific substrates of several non-receptor PTPs have been identified successfully in the past using substrate-trapping mutants (17-19, 24, 25, 47). We used a similar substrate trapping approach to identify physiological substrates of RPTP␣. P19 EC cells were treated with pervanadate, which induced a strong increase in Tyr(P) content of many proteins (Fig. 1A, left two lanes). Bacterially expressed GST fusion proteins containing either the N-termi-nal or the C-terminal catalytic domain of RPTP␣ (D1 or D2, respectively) or mutants thereof were purified and incubated with lysates from pervanadate-treated P19 EC cells. Proteins from the lysate that associated with the fusion proteins were then analyzed by immunoblotting with anti-Tyr(P) mAb (PY20) (Fig. 1). A major Tyr(P)-containing protein of approximately 130 kDa (p130) bound specifically to GST-RPTP␣-D1 with the catalytic site Cys mutated to Ser (D1-C433S). p130 is the most prominent phosphotyrosyl protein in the lysate from pervanadate-treated P19 EC cells. Nevertheless, it is clear that RPTP␣-C433S bound preferentially to p130 (Fig. 1A, and not shown overexposures of the blot in Fig. 1A). No tyrosine-phosphorylated proteins were detected to bind to wild type GST-RPTP␣-D1 or GST alone (Fig. 1A).
Substrate-trapping mutants of PTP1B and PTP-PEST in which the general base/general acid Asp was mutated to Ala were shown to be even more efficient substrate-trapping mutants than the catalytic site Cys mutants (17,18). RPTP␣-D1 with the corresponding Asp to Ala mutation, GST-D1-D401A, did not bind any Tyr(P)-containing proteins from the lysate (Fig. 1A). This suggests that there are differences in substrate binding between the Asp mutants in PTP1B or PTP-PEST and RPTP␣-D1. It has been shown for PTP1B that mutation of the invariant Arg in the signature motif, Arg 221 in PTP1B, almost completely abolished catalytic activity (17). Similarly, GST-RPTP␣-D1 with the corresponding mutation, D1-R438K, had no detectable PTP activity in in vitro PTP assays toward any of the substrates tested (data not shown). No tyrosine-phosphorylated proteins were detected to bind to GST-D1-R438K (Fig.  1A). Since p130 bound specifically to GST-D1-C433S and not to D1-D401A or D1-R438K, we conclude that p130 is not merely a RPTP␣-binding protein but a bona fide substrate.
The apparent molecular weight of the protein that bound to RPTP␣-D1-C433S prompted us to test whether this protein was p130 cas . The blot depicted in Fig. 1A was stripped and reprobed with anti-p130 cas antibody (Fig. 1A, bottom panel). The substrate-trapping mutant GST-D1-RPTP␣-C433S bound to a single protein detected by the anti-p130 cas antibody. This protein co-migrated exactly with p130 cas in lysates from pervanadate-treated cells. p130 cas in lysates from untreated cells migrated faster in the gel than p130 cas in lysates from pervanadate-treated cells (Fig. 1A, 1st two lanes, bottom panel), consistent with reports that phosphorylation of p130 cas induced a mobility shift in SDS-PAGE gels (48). p130 cas did not bind to wild type RPTP␣-D1, D1-D401A, D1-R438K, or GST alone, demonstrating that p130 cas binding is specific for the D1-C433S substrate-trapping mutant. p130 cas did not bind to GST-D2 or any of its mutants either. In conclusion, p130 from pervanadate-treated P19 EC cells that bound to GST-D1-C433S is p130 cas , and p130 cas bound specifically to the substrate-trapping mutant RPTP␣-D1-C433S.
Aliquots of the remaining P19 EC cell lysates were analyzed by immunoblotting using anti-Tyr(P) antibody PY20 following the pull-down procedure, using the different GST fusion proteins (Fig. 1B). Incubation of the lysate with GST-RPTP␣-D1-WT reduced the Tyr(P) content of all the Tyr(P)-containing proteins in the lysate, as detected with anti-Tyr(P) antibody, PY20. Importantly, phosphorylation of the highly phosphorylated p130 was greatly reduced. Incubation with GST or the other GST fusion proteins had no effect on the Tyr(P) levels of the proteins in the lysate, except for p130 in lysate incubated with GST-D1-C433S that showed a significant reduction. Stripping the blot and reprobing it with anti-p130 cas antibodies showed that the highly phosphorylated 130-kDa protein was p130 cas (Fig. 1B, bottom panel). The amount of p130 cas in the lysate was reduced after pull down with GST-RPTP␣-D1-C433S, which was consistent with the removal of p130 cas from the lysate by the pull-down procedure. Incubation of the lysate from pervanadate-treated P19 EC cells with GST-RPTP␣-D1-WT induced a shift down of all p130 cas (Fig. 1B, bottom  panel), suggesting that none of the slower migrating tyrosinephosphorylated p130 cas remained, and thus that p130 cas tyrosine dephosphorylation was complete. p130 cas did not shift down to the same level as p130 cas from untreated cells, which is probably due to residual serine/threonine phosphorylation. In conclusion, GST-RPTP␣-D1-WT preferentially dephosphorylated p130 cas , suggesting that dephosphorylation of p130 cas by RPTP␣ is specific.
Similar results were obtained using NIH3T3 cells instead of P19 EC cells. GST-RPTP␣-D1-C433S specifically bound to a 130-kDa protein that was identified as p130 cas (Fig. 1C). In addition, GST-RPTP␣-D1-WT preferentially dephosphorylated p130 cas from pervanadate-treated NIH3T3 cell lysates (Fig.  1D). In addition to p130 cas , GST-RPTP␣-D1-C433S bound three other phosphotyrosyl proteins from NIH3T3 cells, a doublet of approximately 110 kDa and a protein of 85 kDa. The 110-kDa proteins did not bind to any of the other GST fusion proteins, suggesting that the 110-kDa protein is a bona fide substrate for RPTP␣ too. The 85-kDa protein, however, bound to GST-RPTP␣-D1-D401A and -R438K with similar efficiencies as to -C433S, suggesting that the 85-kDa protein bound to RPTP␣-D1, independently of the catalytic site. In conclusion, tyrosinephosphorylated p130 cas from P19 EC cells and NIH3T3 cells specifically bound to RPTP␣-D1-C433S.
Interaction of RPTP␣-D1-C433S with p130 cas from Different Cell Types-To determine whether p130 cas binding to RPTP␣-D1-C433S was cell type-specific, we incubated GST-RPTP␣-D1-C433S with cell lysates from a range of different cells treated with pervanadate (Fig. 2). GST-RPTP␣-D1-C433S bound to p130 cas from all cell types we tested. Furthermore, as seen for NIH3T3 cells, GST-RPTP␣-D1-C433S bound three additional proteins of 85 and 110 kDa from F9, FaO, and HepG2 cells. GST-RPTP␣-D1-C433S bound an additional phosphotyrosyl protein from HepG2 cell lysate of about 70 kDa and one from FaO cell lysate of about 80 kDa that were not seen in the other cells. Long exposures demonstrated that little or no Tyr(P)-containing proteins were detected to bind to GST-RPTP␣-D1-C433S from cells that were not treated with pervanadate ( Fig. 2 and data not shown). Both GST-RPTP␣-D1-WT and GST alone failed to bind to any tyrosinephosphorylated protein from the different cell lysates (data not shown). In conclusion, RPTP␣-D1-C433S associated with tyrosine-phosphorylated p130 cas from all cell types tested.
Specific Interaction of RPTP␣-D1 with p130 cas -Next, we compared the specificities of RPTP␣-D1 and PTP1B. GST fusion proteins of RPTP␣-D1 and of wild type PTP1B and PTP1B-C213S were incubated with lysate from pervanadate-treated P19 EC cells. GST-RPTP␣-D1-C433S bound predominantly one protein from P19 EC cell lysate (Fig. 3). As reported before (17) the pattern of the Tyr(P)-containing proteins bound by the PTP1B substrate-trapping mutant strongly resembled that of the input lysate, indicating that the PTP1B substrate-trapping mutant bound most phosphotyrosyl proteins from the lysate with similar efficiencies. No tyrosine-phosphorylated proteins were detected to bind to the wild type fusion proteins of RPTP␣-D1 and PTP1B. In conclusion, the PTP1B substratetrapping mutant showed little or no substrate specificity, whereas RPTP␣-D1-C433S bound predominantly to one protein from P19 EC cells.  Binding of p130 cas from Fibronectin-stimulated Cells to GST-RPTP␣-D1-C433S-Pervanadate treatment of cells induced massive tyrosine phosphorylation of many proteins, including p130 cas . In order to investigate whether GST-RPTP␣-D1-C433S recognized p130 cas under physiological conditions, we stimulated NIH3T3 fibroblasts by plating them on fibronectin (FN), a well known stimulus for p130 cas tyrosine phosphorylation (49 -51). Hardly any p130 cas tyrosine phosphorylation was detected in cells that were detached from the substratum for 30 min, and replating of the cells on FN rapidly led to recovery of p130 cas phosphorylation (Fig. 4), consistent with previous reports (37). GST-RPTP␣-D1-C433S pull-down experiments showed that tyrosine-phosphorylated p130 cas from FN-stimulated cells, but not unphosphorylated p130 cas from detached cells, was recognized by RPTP␣-D1-C433S (Fig. 4). Long exposures are depicted in Fig. 4, as compared with Figs. 1 and 2, leading to detection of p130 cas from control cells as well. Pull down of p130 cas from pervanadate-treated cells was much more efficient, due to much higher levels of p130 cas tyrosine phosphorylation. These results demonstrate that not only pervanadate-but also fibronectin-stimulated tyrosine-phosphorylated p130 cas bound to RPTP␣-D1-C433S, suggesting that p130 cas is a physiological substrate of RPTP␣.
p130 cas Is a Substrate of RPTP␣ in Vitro-We tested the capacity of RPTP␣ to dephosphorylate p130 cas in vitro. As a source of tyrosine-phosphorylated p130 cas , we used transiently transfected SK-N-MC cells co-expressing GST-p130 cas and SrcY529F. Tyrosine-phosphorylated GST-p130 cas was purified from these cells, pooled, aliquoted, and incubated with a limited amount of purified bacterially expressed RPTP␣ for different periods. The Tyr(P) content of p130 cas was analyzed by immunoblotting with an anti-Tyr(P) mAb (PY20). Incubation of p130 cas with bacterially expressed RPTP␣ resulted in complete dephosphorylation of p130 cas , which was accompanied by a shift down of p130 cas , due to dephosphorylation (Fig. 5). p130 cas did not shift down to prestimulation levels, which was presumably due to residual serine/threonine phosphorylation. Incubation of p130 cas without RPTP␣ had no effect on the phosphorylation state of p130 cas (Fig. 5). In conclusion, RPTP␣ completely dephosphorylated p130 cas in vitro.
p130 cas and RPTP␣ Co-localized at the Membrane-For RPTP␣ to dephosphorylate p130 cas in vivo, both proteins must co-localize subcellularly. To examine subcellular localization of RPTP␣ and p130 cas , SK-N-MC cells, plated on glass coverslips, were transiently transfected with RPTP␣ in which D2 was replaced by yellow fluorescent protein (YFP). p130 cas was visualized by indirect immunofluorescence using anti-p130 cas antibodies. Confocal microscopy demonstrated that p130 cas localized to the cytoplasm and to the plasma membrane (Fig. 6). RPTP␣-YFP localized predominantly to the plasma membrane. The combined image of RPTP␣-YFP and p130 cas demonstrated that p130 cas and RPTP␣ co-localized at the membrane. These results show that RPTP␣ co-localized with a subpopulation of p130 cas , suggesting that RPTP␣ is in the right subcellular location to dephosphorylate p130 cas .
p130 cas Is a Substrate of RPTP␣ in Vivo-Finally, we investigated whether p130 cas was an in vivo substrate of RPTP␣. SK-N-MC cells were transiently transfected with GST-p130 cas together with expression vector, wild type HA-RPTP␣, or HA-RPTP␣-C433S. The cells were co-transfected with active Src (SrcY529F), strongly increasing tyrosine phosphorylation of GST-p130 cas (Fig. 7A). Co-expression of wild type RPTP␣ in these cells significantly reduced tyrosine phosphorylation of p130 cas , which was accompanied by a shift in migration (Fig.  7A). Basal level p130 cas tyrosine phosphorylation in cells that were not co-transfected with active Src was not affected significantly by co-expression of RPTP␣. Analysis of the Tyr(P) content of proteins in the lysates of cells co-transfected with SrcY529F and RPTP␣ indicated that not all proteins were dephosphorylated upon co-transfection of RPTP␣ (Fig. 7A, bot-FIG. 4. RPTP␣-D1-C433S substrate trapping of fibronectinstimulated p130 cas . Serum-starved NIH3T3 cells were left untreated (On), taken off the dish with EDTA (Off), or replated on fibronectincoated (FN) dishes for 30 min as described under "Materials and Methods." As a positive control, NIH3T3 cells were stimulated for 30 min with 1 mM orthovanadate and 1 mM H 2 O 2 . The cells were lysed, and pull downs were performed with GST-RPTP␣-D1-C433S. Putative substrates were analyzed by SDS-PAGE and immunoblotting, using anti-Tyr(P) mAb PY20 (top panel). To establish that tyrosine-phosphorylated p130 in the top panel was p130 cas , proteins were eluted from the glutathione-agarose beads following the pull down and immunoprecipitated (IP) using anti-Tyr(P) mAb PY20. These immunoprecipitates were analyzed by SDS-PAGE and immunoblotting, using anti-p130 cas antibodies (second panel). Total cell lysates were analyzed in parallel to monitor Tyr(P) (3rd panel) and p130 cas levels (bottom panel). Exposure times were 30 min for the anti-Tyr(P) blots, except for the vanadate lanes (30 s), 2 min for the p130 cas blot of the pull downs, except for the vanadate lane (30 s), and 30 s for the p130 cas blot of the cell lysates. Note the shift in p130 cas in vanadate-treated cells. Highly tyrosinephosphorylated p130 cas is poorly recognized by the anti-p130 cas antibody (bottom panel) (cf. Fig. 5).
FIG . 5. p130 cas dephosphorylation by RPTP␣ in vitro. SK-N-MC cells were transiently transfected with GST-p130 cas together with SrcY529F. The cells were lysed, and GST-p130 cas was purified, aliquoted, and incubated with purified bacterially expressed PTP␣ encoding the complete cytoplasmic domain of RPTP␣ (bPTP␣) for 5, 10, 15, 20, or 30 min, or with GST (Ϫ) for 30 min at 30°C. The samples were analyzed by SDS-PAGE and immunoblotting with anti-Tyr(P) antibodies (PY20) (top panel) or anti-p130 cas antibodies (bottom panel). Note that tyrosine-phosphorylated p130 cas is poorly recognized by the anti-p130 cas antibody (cf. lane 1 and 6). tom panel), suggesting that RPTP␣ did not merely dephosphorylate all tyrosine-phosphorylated proteins. As a control, we investigated whether RPTP␣ dephosphorylated tyrosine-phosphorylated EGFR. The EGFR was co-transfected with control vector, HA-RPTP␣, or HA-RPTP␣-C433S, and the cells were stimulated with EGF or left untreated. Co-transfection of RPTP␣ did not reduce EGF-induced EGFR tyrosine phosphorylation (Fig. 7B, cf. wt and CS lanes). Overexpression of the EGFR at high levels led to ligand-independent basal level tyrosine phosphorylation of the EGFR, which was not reduced by co-transfection of RPTP␣ either. The apparent increase in EGFR tyrosine phosphorylation upon co-transfection of (mutant) HA-RPTP␣ was due to differences in expression levels (Fig. 7B, cf. top and middle panels). Taken together, RPTP␣ co-expression in SrcY529F-expressing cells led to a reduction in p130 cas tyrosine phosphorylation, strongly suggesting that RPTP␣ dephosphorylated p130 cas in vivo. Moreover, RPTP␣ co-expression did not reduce EGFR tyrosine phosphorylation, suggesting that RPTP␣ displays substrate specificity in living cells.

DISCUSSION
In order to understand the function of PTPs, it is essential to know the identity of their substrates. We have used a substrate trapping procedure to identify substrates of RPTP␣, a transmembrane PTP with a short extracellular domain. RPTP␣-D1-C433S bound specifically to tyrosine-phosphorylated proteins from pervanadate-treated cells. One of these proteins was identified as p130 cas , and RPTP␣-D1-C433S bound tyrosine-phosphorylated p130 cas from fibronectin-stimulated cells as well, suggesting that p130 cas is a physiological substrate of RPTP␣. RPTP␣-D1 preferentially dephosphorylated p130 cas from pervanadate-treated P19 EC and NIH3T3 cell lysates (Fig. 1, B  and D). This preference was surprising, since previously we demonstrated that RPTP␣ displayed only modest selectivity toward peptide substrates in vitro (42). In living cells, p130 cas , but not EGFR tyrosine phosphorylation, was reduced upon co-transfection of RPTP␣ (Fig. 7). These results suggest that RPTP␣ displays substrate specificity in vivo.
Substrate trapping experiments with PTP-PEST and PTP1B demonstrated that a substrate-trapping mutant, in which the Asp residue that functions as general base in catalysis was mutated to Ala, displayed much higher affinity for Tyr(P)containing proteins than catalytic site Cys to Ser mutants (18). . The cells were co-transfected with the expression vector pSG-5 (Ϫ) or SrcY529F. The cells were lysed, and GST-p130 cas was precipitated and analyzed by SDS-PAGE and immunoblotting with anti-Tyr(P) mAb (PY20) or anti-p130 cas antibodies. The left part of the anti-p130 cas blot (2nd panel from the top) was exposed for 10 s, and the right part was exposed for 2 min, since tyrosine phosphorylation reduced detection of p130 cas tremendously (cf. Fig. 5). HA-RPTP␣ protein expression levels were monitored by immunoprecipitation with 12CA5 mAb and immunoblotting with 12CA5 mAb (3rd panel). Tyr(P) levels in the lysates were determined by immunoblotting, using anti-Tyr(P) mAb, PY20. A part of the gel (between 66 and 97 kDa) is shown in the bottom panel, and Tyr(P)-containing bands that are not affected by RPTP␣ are indicated with black arrowheads, whereas proteins that are less phosphorylated upon co-transfection of RPTP␣, like p130 cas , are indicated with open arrowheads. B, SK-N-MC cells were transiently transfected with an expression vector for the human EGFR, together with expression vector pSG-5 (Ϫ), pSG-HA-RPTP␣ (wt), or pSG-HA-RPTP␣-C433S (CS). The cells were stimulated with EGF (10 ng/ml) for 5 min, or left untreated. The EGFR was immunoprecipitated and analyzed by anti-Tyr(P) and anti-EGFR immunoblotting (top two panels). SK-N-MC cells do not express endogenous EGFR (middle panel). The expression of co-transfected (mutant) RPTP␣ was monitored by immunoblotting, using an anti-RPTP␣ antibody. RPTP␣-D1 with the similar Asp to Ala mutation (RPTP␣-D401A) did not bind any Tyr(P)-containing protein from pervanadate-treated P19 EC cells or NIH3T3 cells (Fig. 1), demonstrating that the WpD motif Asp mutants are not always the best substrate-trapping mutants.
p130 cas is not the only substrate of RPTP␣, since several other tyrosine-phosphorylated proteins from different cell lines bound to RPTP␣-D1-C433S. For instance, RPTP␣-D1-C433S bound, next to p130 cas , three other proteins, a doublet of 110 kDa and a protein of 85 kDa from NIH3T3 cells. The p85 protein bound to RPTP␣-D401A and RPTP␣-D1-R438K as well, suggesting that binding to p85 was independent of the catalytic site. p85 was not detected to bind to wild type RPTP␣-D1, which may be due to dephosphorylation of p85 by RPTP␣-D1. Perhaps p85 is not only a binding protein but also a substrate of RPTP␣. Identification of p85 will facilitate rigorous testing of the interaction between p85 and RPTP␣. The p110 protein specifically bound to RPTP␣-D1-C433S and not to any of the other GST fusion proteins, indicating that p110 is an additional RPTP␣ substrate. The apparent molecular weight of p110 and p85 suggested that these proteins might be phosphatidylinositol 3-kinase. However, antibodies directed at the p110 subunit of phosphatidylinositol 3-kinase did not recognize the 110-kDa substrate of RPTP␣ (data not shown), suggesting phosphatidylinositol 3-kinase is not a substrate of RPTP␣. Currently, we are trying to identify p110 and p85 by protein purification and microsequencing.
Other substrates previously described for RPTP␣ include the related PTKs Src and Fyn. RPTP␣ activates Src in vitro and in vivo by dephosphorylating the inhibiting Tyr(P) 529 (11, 30 -32). Rat embryo fibroblasts, P19 EC cells, and A431 cells overexpressing RPTP␣ showed an increase in Src activity (11,30,36). In RPTP␣ Ϫ/Ϫ cells Src and Fyn activity is reduced (31), providing strong support that these PTKs are substrates of RPTP␣. By using the substrate trapping experiments, we did not pull down proteins with apparent molecular weights that correspond to Src and Fyn. This may be due to poor binding of RPTP␣-D1-C433S to Src and Fyn. Moreover, failure to detect Src and Fyn in RPTP␣-D1-C433S pull downs may be due to poor recognition of the C-terminal regulatory Tyr(P) in Src and Fyn by the anti-Tyr(P) antibody, PY20.
For all RPTPs with two PTP domains, the majority of the catalytic activity resides within D1, whereas D2 displays little or no catalytic activity. D2s are thought to have a more regulatory role. The existence of naturally occurring inactive PTP domains, such as D2s, has led to the suggestion that RPTP-D2s function as Tyr(P)-binding modules (52). Here we demonstrate that GST-RPTP␣-D2 did not bind any Tyr(P)-containing proteins from pervanadate-treated cells (Fig. 1). GST-RPTP␣-D2 with the catalytic site Cys mutated to Ser (C723S) or another inactive D2 mutant (R729K) did not bind Tyr(P)-containing proteins either. RPTP␣-D2 lacks the highly conserved Tyr in the KNRY motif (14) that interacts with the main chain atoms and the aromatic ring of the substrate Tyr(P) (53). Introduction of this conserved Tyr in RPTP␣-D2 (V555Y) did not restore Tyr(P) binding (data not shown). We conclude that it is unlikely that RPTP␣-D2 by itself is a Tyr(P)-binding module. However, we cannot exclude the possibility that RPTP␣-D2 cooperates with RPTP␣-D1 in binding to substrates.
Other PTPs, like PTP1B, PTP-PEST, and LAR, have been found to dephosphorylate p130 cas as well (18,54,55). Both PTP1B and PTP-PEST have been found to associate with and dephosphorylate p130 cas (18,54,56,57). Interaction of these PTPs with p130 cas is mediated by a proline-rich region in PTP1B or PTP-PEST and the SH3 domain of p130 cas (54,56,58). The p130 cas SH3 domain-PTP-PEST interaction is not required for the interaction of PTP-PEST with p130 cas , since substrate-trapping mutants of the PTP-PEST catalytic domain bound to p130 cas in the absence of the proline-rich region (18). RPTP␣ contains a proline-rich region (PPLP, residues 210 -213), but this region is not sufficient for p130 cas binding, since p130 cas did not bind to wild type RPTP␣ nor the inactive mutants D401A and R438K that all contained the proline-rich region (Fig. 1).
p130 cas is part of focal adhesion complexes. Turnover of focal adhesion complexes is essential for cell movement and outgrowth of cell extensions. Interfering with PTP1B, PTP-PEST, or RPTP␣ resulted in altered cell motility (18, 32, 58 -61), suggesting that these PTPs are somehow involved in regulating signaling of focal adhesion complexes. Regulation of p130 cas tyrosine phosphorylation by these PTPs may be the mechanism that underlies involvement of these PTPs in cell motility.
Here, we demonstrate that tyrosine-phosphorylated p130 cas is a substrate of RPTP␣. Activated SrcY529F-induced tyrosine phosphorylation of p130 cas is clearly reduced in cells that express active RPTP␣ (Fig. 7). Src is negatively regulated by phosphorylation in its C terminus (Tyr 529 ) and positively regulated by phosphorylation of Tyr 416 . RPTP␣ dephosphorylates both Tyr(P) 529 and Tyr(P) 416 in Src (11). We used an excess of SrcY529F in our experiments, and therefore the effects of RPTP␣ on SrcY529F activity via dephosphorylation of Tyr(P) 416 are negligible, which is illustrated by the fact that co-expression of RPTP␣ did not affect tyrosine phosphorylation of all proteins in the lysate (Fig. 7A, bottom panel). Therefore, the effect of RPTP␣ on p130 cas tyrosine phosphorylation is due to dephosphorylation of p130 cas , not to reduced activity of SrcY529F.
The interaction between RPTP␣ and p130 cas may be complex. Src and Fyn are substrates of RPTP␣, and these PTKs are activated by dephosphorylation of the C-terminal regulatory Tyr(P). Active Src and Fyn in turn phosphorylate p130 cas (48,62). Therefore, RPTP␣ may have a dual effect on p130 cas tyrosine phosphorylation. RPTP␣ may dephosphorylate p130 cas directly, and RPTP␣ may induce enhanced phosphorylation of p130 cas via activation of Src and Fyn. Overexpression of RPTP␣ in the absence of activated SrcY529F did not affect p130 cas tyrosine phosphorylation significantly (Fig. 6), which may be due to the dual effect of RPTP␣.
In conclusion, we provide evidence that RPTP␣ dephosphorylatyed p130 cas in vitro and in vivo. p130 cas contains many putative tyrosine phosphorylation sites. It remains to be determined whether all of these sites are substrates for RPTP␣ or for any other PTP. Different PTPs may dephosphorylate different Tyr(P) sites in p130 cas . We are currently investigating this interesting possibility.