Mitotic activation of protein-tyrosine phosphatase alpha and regulation of its Src-mediated transforming activity by its sites of protein kinase C phosphorylation.

During mitosis, the catalytic activity of protein-tyrosine phosphatase (PTP) alpha is enhanced, and its inhibitory binding to Grb2, which specifically blocks Src dephosphorylation, is decreased. These effects act synergistically to activate Src in mitosis. We show here that these effects are abrogated by mutation of Ser180 and/or Ser204, the sites of protein kinase C-mediated phosphorylation within PTPalpha. Moreover, either a Ser-to-Ala substitution or serine dephosphorylation specifically eliminated the ability of PTPalpha to dephosphorylate and activate Src even during interphase. This explains why the substitutions eliminated PTPalpha transforming activity, even though PTPalpha interphase dephosphorylation of nonspecific substrates was only slightly decreased. This occurred without change in the phosphorylation of PTPalpha at Tyr789, which is required for "phosphotyrosine displacement" during Src dephosphorylation. Thus, in addition to increasing PTPalpha nonspecific catalytic activity, Ser180 and Ser204 phosphorylation (along with Tyr789 phosphorylation) regulates PTPalpha substrate specificity. This involves serine phosphorylation-dependent differential modulation of the affinity of Tyr(P)789 for the Src and Grb2 SH2 domains. The results suggest that protein kinase C may participate in the mitotic activation of PTPalpha and Src and that there are intramolecular interactions between the PTPalpha C-terminal and membrane-proximal regions that are regulated, at least in part, by serine phosphorylation.

PTP␣ down-regulation (10), indicating that PTP␣ is a major physiological positive regulator of Src.
Constitutive activation of the Src proto-oncoprotein by mutation increases the tyrosine phosphorylation of multiple signal transduction proteins and thereby neoplastically transforms a variety of cell types (see Ref. 7 for review). The fact that activation by overexpressed PTP␣ also causes transformation (3) is perhaps more surprising and suggests that PTP␣ activity is directed in vivo preferentially to Src (and Src family members), rather than to Src substrates.
This substrate specificity is due, at least in part, to a phosphotyrosine displacement mechanism that selectively promotes dephosphorylation of Src by PTP␣: ϳ20% of PTP␣ in NIH3T3 cells is phosphorylated at Tyr 789 , a residue near its carboxyl terminus (11,12). Tyr 789 phosphorylation does not affect PTP␣ dephosphorylation of nonspecific substrates such as myelin basic protein (MBP), whose phosphotyrosines are not bound, but is required for dephosphorylation of Src Tyr(P) 527 , which is protected against many phosphatases by its SH2 domain binding (13). Phosphorylated Tyr 789 can bind to the Src SH2 domain, thereby displacing and thus unprotecting Src Tyr(P) 527 . This also forms a transient bound state that additionally facilitates Tyr(P) 527 dephosphorylation (13).
Tyr(P) 789 also binds the SH2 domain of the adapter protein Grb2 (11,12), which participates in Ras activation following peptide growth factor stimulation (see Ref. 14 for review). Because of steric hindrance resulting from the interaction of one of the Grb2 SH3 domains with PTP␣, PTP␣-bound Grb2 is not able to bind Sos, the downstream protein in the Grb2-Ras signal transduction pathway. Thus, it does not appear that localization of Grb2 to the plasma membrane by binding to PTP␣ can activate the Ras signaling pathway (15, 16). Instead, control may flow in the other direction: Grb2 binding to Tyr(P) 789 blocks phosphotyrosine displacement and the ability of PTP␣ to dephosphorylate Src, so only Grb2-unbound, Tyr(P) 789 -phosphorylated PTP␣ is able to activate Src (13). Most Tyr(P) 789 -phosphorylated PTP␣ is bound by Grb2 (11), so small changes in Grb2 binding can sensitively control Srcdirected PTP␣ activity.
The mitotic activation of PTP␣ has two components: 1) its catalytic activity, as measured on nonphysiological substrates such as MBP, increases ϳ2-fold; and 2) the inhibitory binding of Grb2 to PTP␣ is reduced 3-4-fold (24). The latter reduction occurs because of a mitotic decrease in the affinity of PTP␣ for the Grb2 SH2 domain without a decrease in its affinity for the Src SH2 domain. This results in 2-3-fold increased Src-PTP␣ co-association and a commensurate increase in Src-directed PTP␣ activity (24). This relief from Grb2 competition combines multiplicatively with the increase in catalytic activity to give a 4 -5-fold increase in total Src-directed PTP␣ activity (24).
The mechanism(s) responsible for increasing PTP␣ specific activity and selectively decreasing its binding to the Grb2 SH2 domain are not known. The changes occur without altered Tyr 789 phosphorylation and can be observed with purified PTP␣ in vitro. Moreover, there is no evidence of PTP␣ dimerization under the experimental conditions. The changes coincide with mitotic reduction of PTP␣ electrophoretic mobility, suggesting that hyperphosphorylation is involved. Indeed, treating PTP␣ with the Ser/Thr-specific phosphatase PP2A coordinately eliminates PTP␣ mitotic mobility retardation, increased catalytic activity, and decreased Grb2 binding. The effect on Src-directed PTP␣ activity is even stronger: PP2A treatment not only blocks the mitotic increase, but also reduces, if not eliminates, the ability of interphase PTP␣ to activate Src (24). Because PTP␣ is predominantly phosphorylated at serine and has very little or no threonine phosphorylation (24 -26), this suggests that the mitotic activation of PTP␣ requires, and may be caused by, serine hyperphosphorylation (24).
Two serine phosphorylation sites in PTP␣ have already been identified: Ser 180 and Ser 204 can both be phosphorylated in vitro in NIH3T3 cells by protein kinase C and are phosphorylated in vivo following treatment of cells with 12-O-tetradecanoylphorbol-13-acetate (25, 26). 12-O-Tetradecanoylphorbol-13-acetate-induced serine hyperphosphorylation of PTP␣ increases its catalytic activity 2-3-fold (25), probably because of phosphorylation at these sites (26). Recently, it was shown that phosphorylation at Ser 180 and Ser 204 is required for the activation of PTP␣ that follows treatment of A431 cells with a somatostatin analog (27).
To investigate the possibility that phosphorylation at Ser 180 and/or Ser 204 is involved in the mitotic activation of PTP␣, we have compared the effects of separate and coordinate sitespecific substitutions at these sites with those of PP2A treatment. We found that these phosphorylations are required for mitotic activation of PTP␣, for the change in its SH2 domainbinding properties, and for its ability to activate Src and to transform cells.

MATERIALS AND METHODS
Antibodies-All PTP␣ immunoprecipitations and immunoblotting were performed with polyclonal antibody 7-091, which was made in rabbits against a GST fusion protein containing PTP␣ residues 165-793 (13). Anti-HA immunoprecipitations were carried out with monoclonal antibody 12CA5 (28).
Cell Lines, Nocodazole Arrest of Mitotic Cells, and Induction of PTP␣ Expression-Except for cell lines overexpressing the S180A and S204A mutants (described below), all lines were previously described (13). Cells were grown; PTP␣ expression was induced; and cells were arrested in mitosis with nocodazole and collected by mechanical shake-off as described (24).
Mutant PTP␣-inducible Expression Plasmids and Cell Lines-Plasmids for inducible expression of the Ser-to-Ala human PTP␣ mutants were constructed by replacing coding sequences lying between the two HindIII sites in WT PTP␣ expression plasmids pNTPTP␣ (no HA tag) and pTPTP␣ (with the HA epitope tag YPYDVPDYA) with mutated sequences constructed by PCR. PCR products and restriction fragments that together comprised complete (mutated) coding sequences lying between the HindIII sites were then religated into vector plasmid pTet-Splice (Invitrogen) to construct plasmids that were identical to pNTPTP␣ or pTPTP␣ except for the specified mutations. For the S180A substitution, a mutated HindIII-EclXI fragment was prepared by PCR using pNTPTP␣ as a template with the 5Ј-primer 5Ј-CGCCAAGCTTG-GCCACCATGGATTCCTGGTTCATTCTTGT-3Ј and the 3Ј-primer 5Ј-CAGTGCGGCCGTTGGATAAGCGGAAAGCATTGGAAT-3Ј. The 5Јprimer contained the HindIII site (underlined) and the start codon (boldface); the 3Ј-primer contained the EclXI site (underlined). The substitution AGA 3 AGC (boldface italics) in the 3Ј-primer resulted in the S180A substitution. This PCR product was cleaved with HindIII and EclXI, mixed with the complementary gel-purified EclXI-HindIII restriction fragment from pNTPTP␣, and ligated into the HindIII site of pTet-Splice to make plasmid pNTPTP␣(S180A). Plasmid pTPTP␣(S180A), which expresses PTP␣(S180A)-HA, was constructed similarly, except that the EclXI-HindIII fragment was from pTPTP␣.
For the S204A mutation, a 0.6-kilobase pair HindIII-BglII fragment containing the 5Ј-portion of the WT coding sequence was copied from pNTPTP␣ by PCR using the 5Ј-primer 5Ј-CGCCAAGCTTGGCCAC-CATGGATTCCTGGTTCATTCTTGT-3Ј and the 3Ј-primer 5Ј-TTGGT-GGCTGGAGATCTGGCCAGAAGTGGCACACTCT-3Ј. The 5Ј-primer contained the HindIII site (underlined) and the start codon (boldface); the 3Ј-primer contained the BglII site (underlined). This PCR product was cleaved with HindIII and BglII. (Although the 3Ј-primer contained the substitution AGC 3 GGC at nucleotides 6 -8, the BglII digestion removed this region.) The BglII-HindIII fragment containing the downstream coding sequence with the S204A mutation was generated using pNTPTP␣ as a PCR template with the 5Ј-primer 5Ј-TGGCCAGATCTC-CAGCCACCAACAGGAAATACCCACCCCT-3Ј and the 3Ј-primer 5Ј-TGTTGAAGCTTACTTGAAGTTGGCATAATC-3Ј. The 5Ј-primer contained the BglII site (underlined); the 3Ј-primer contained the stop codon (boldface) and the HindIII site (underlined). The mutation AGC 3 GCC (boldface italics) in the 5Ј-primer caused the S204A substitution. This PCR product was cleaved with BglII and HindIII, and both fragments were ligated into the HindIII site of pTet-Splice to make plasmid pNTPTP␣(S204A). Plasmid pTPTP␣(S204A), which expresses PTP␣(S204A)-HA, was constructed by ligating the gel-purified HindIII-ClaI restriction fragment from pNTPTP␣(S204A), containing the S204A mutation, along with the complementary gel-purified ClaI-HindIII restriction fragment from pTPTP␣, containing the downstream WT PTP␣-HA sequence, into the HindIII site of pTet-Splice.
For the S202A mutation, an EclXI-HindIII fragment containing the mutation and the downstream coding sequence was prepared by PCR using pNTPTP␣ as a template with the 5Ј-primer 5Ј-CCAACGGCCGCACTGAGGATGTGGAGCCCCAGAGTGTGCCA-CTTCTGGCCAGAGCCCCA-3Ј and the 3Ј-primer 5Ј-TGTTGAAGCT-TACTTGAAGTTGGCATAATC-3Ј. The 5Ј-primer contained the EclXI site (underlined); the 3Ј-primer contained the stop codon (boldface) and the HindIII site (underlined). The mutation TCC 3 GCC (boldface italics) in the 5Ј-primer resulted in the S202A substitution. This PCR product was digested with EclXI and HindIII, mixed with the 0.57-kilobase pair gel-purified HindIII-EclXI restriction fragment from pNTPTP␣, and ligated into the HindIII site of pTet-Splice to make plasmid pNTPTP␣(S202A).
The mutations were verified by sequencing of the PTP␣ coding region. These plasmids were stably cotransfected with the G418 resistance plasmid pSV2neo (29) and the tetracycline transactivator plasmid pTet-tTak (Invitrogen) into NIH3T3 cells, selected for G418 resistance and for inducible expression of the PTP␣ mutants as described (13).
Immunoprecipitation and Immunopurification of PTP␣, Dephosphorylation and Kinase Assays, Co-immunoprecipitation and Affinity Precipitation Assays, and PP2A Serine Dephosphorylation-These were performed as described previously (24).
Anchorage-independent Growth Assay-Cells were assayed for colony formation on 0.3% agarose without doxycycline as described previously (30).

RESULTS
We have previously described genetically modified NIH3T3 cell lines that inducibly overexpress (under repressive control of doxycycline) human WT PTP␣ and PTP␣(Y789F) and the same proteins with a nine-residue HA tag at their C termini, designated PTP␣-HA and PTP␣(Y789F)-HA (13). A "Neo" cell line that had been transfected with an empty vector system and co-selected in the same manner provided a control for analyzing endogenous PTP␣. It was previously shown that the localization and specific catalytic activity of overexpressed WT PTP␣ are similar to those of endogenous PTP␣ in both unsynchronized and mitotic cells (13,24).
New plasmids and corresponding NIH3T3-derived cell lines for inducible expression of PTP␣ (with or without the HA tag) with Ser-to-Ala substitutions at residues 180 and/or 204 were created using similar methods (see "Materials and Methods"). A cell line expressing PTP␣ with a Ser-to-Ala substitution at residue 202 (the only potential site of cyclin-dependent kinase or mitogen-activated protein kinase serine phosphorylation within PTP␣) was also generated as a control for some experiments. The overexpresser cells maximally expressed ϳ10 -20 times the amount of endogenous PTP␣ when grown in the absence of doxycycline for Ն16 h. So that equal levels of overexpression could be obtained in both unsynchronized and mitotic cells, the time of induction was controlled so that transgene PTP␣ expression was induced only to ϳ5-10 times endogenous levels for the biochemical experiments (see Ref. 24 and "Materials and Methods"). Expression levels were similar within each group of cell lines expressing untagged or HAtagged proteins (data not shown).
To see if we could detect serine phosphates added during mitosis, unsynchronized and nocodazole-arrested mitotic WT PTP␣ and PTP␣(S180A/S204A) overexpresser cells were labeled in vivo with [ 32 P]orthophosphate, and the radiolabeled proteins were analyzed by anti-PTP␣ immunoprecipitation, immunoblotting, and autoradiography. To avoid radioactivity-induced G 2 arrest (17), it was necessary to label all the cells for relatively short (2-3 h) periods and the mitotic cells after nocodazole arrest. Thus, equilibrium labeling was probably not achieved, and only phosphorylations that were catalyzed during metaphase (the point of nocodazole arrest) were detected in the mitotic cells. Under these conditions, no significant changes in the stoichiometry of labeling were observed between WT and mutant PTP␣ from unsynchronized or mitotic cells (data not shown). WT PTP␣ and PTP␣(S180A/S204A) from unsynchronized and mitotic cells displayed similar radioactive phosphoamino acid compositions: a significant excess of phosphoserine over phosphotyrosine and very little or no phosphothreonine (data not shown). We conclude that the serine phosphorylation(s) that cause the electrophoretic mobility retardation of mitotic PTP␣ do not turn over during mitosis and that PTP␣ contains at least one site of serine phosphorylation in addition to Ser 180 and Ser 204 .
Mitotic Increase in PTP␣ Phosphatase Activity Is Blocked by Mutation of Ser 180 or Ser 204 -MBP that had been tyrosinephosphorylated with [␥-32 P]ATP (by v-Src) was incubated with WT or mutant PTP␣-HA that had been immunoprecipitated with anti-HA antibody from unsynchronized or nocodazolearrested mitotic overexpresser cells. The immunoprecipitates were washed with 0.5 M NaCl to remove any co-associated proteins. Specific PTP activity was determined by measuring the relative amounts of 32 P released per molecule of PTP␣.
As previously reported (24), the activity of WT PTP␣ from mitotic cells was about twice that of PTP␣ from unsynchronized cells (Fig. 1A), and there was very little change during mitosis in the amount of PTP␣ (Fig. 1C) or the extent of its tyrosine phosphorylation (Fig. 1B). (Tyr 789 is the only phosphorylated tyrosine in PTP␣ during both interphase and mitosis (13,24), so anti-phosphotyrosine immunoblotting specifically detects its phosphorylation.) The increased specific activity correlated with reduced electrophoretic mobility of PTP␣ (Fig.  1C, compare lanes 1 and 2).
Both coordinate and separate S180A and/or S204A muta-FIG. 1. Nonspecific catalytic activity of PTP␣ from unsynchronized and mitotic cells. Approximately equal amounts of overexpressed human WT PTP␣-HA, PTP␣(S180A/S204A)-HA, PTP␣(S180A)-HA, and PTP␣(S204A)-HA were immunoprecipitated with anti-HA antibody from lysates from NIH3T3-derived overexpresser cells that were either unsynchronized (U) or arrested in mitosis (M). Aliquots of immunoprecipitates that had been washed with a high salt buffer were incubated with [ 32 P]Tyr(P)-containing MBP in phosphatase buffer for 30 min at 30°C or subjected to anti-Tyr(P) or anti-PTP␣ immunoblotting. Additional experiments were performed similarly, except that the WT PTP␣ and PTP␣(S180A/S204A) immunoprecipitates were preincubated with PP2A, a Ser/Thr phosphatase, and then washed with phosphatase buffer before incubation with MBP. A, shown is the amount of 32 P released per molecule of PTP␣ after a 30-min incubation normalized to the amount released by overexpressed PTP␣ from unsynchronized cells. Error bars indicate the S.E.M. from four to five experiments. B, shown is the anti-Tyr(P) (Anti-pTyr) immunoblot of the immunoprecipitated PTP␣. C, shown is the anti-PTP␣ immunoblot of the immunoprecipitated PTP␣. D, experiments were performed as described above using immunoprecipitated WT PTP␣-HA or PTP␣(S180A/S204A)-HA, except that the immunoprecipitates were preincubated with (ϩ) or without (Ϫ) PP2A prior to PTP assay with 32 P-labeled MBP. The amount of 32  tions blocked the mitotic increase in PTP activity (Fig. 1). These mutations also caused a small but reproducible 10 -20% decrease in the specific activity of PTP␣ in unsynchronized cells. The mitotic mobility retardation was completely blocked in PTP␣(S180A/S204A), consistent with the hypothesis that the retardation resulted from mitotic phosphorylation at these sites. The mutations did not significantly affect Tyr 789 phosphorylation.
Similar experiments were conducted with immunoprecipitated WT PTP␣ and PTP␣(S180A/S204A) that had been incubated with the Ser/Thr phosphatase PP2A before incubation with MBP. PP2A treatment slightly reduced the specific activity of interphase PTP␣ and eliminated the mitotic increase in activity (Fig. 1D) without affecting Tyr 789 phosphorylation (Fig. 1E). It also restored the electrophoretic mobility of mitotic PTP␣ almost to its interphase level (Fig. 1F). In contrast, it had no observable effect on the activity or mobility of the serine double mutant. This is consistent with the hypothesis that the PP2A effect results from its dephosphorylation of Ser 180 or Ser 204 .
These results suggest that the mitotic activation of PTP␣ nonspecific catalytic activity requires mitotic hyperphosphorylation of both Ser 180 and Ser 204 . The small decrease in the activity of PTP␣ from unsynchronized cells upon mutation or PP2A treatment may reflect the fact that PTP␣ is phosphorylated to some extent at these sites even during interphase (26).

Tyr 527 Dephosphorylation and Activation of Src in Vitro and in Vivo by PTP␣ Are Blocked by Mutation of Ser 180 or Ser 204 -
The ability of WT and mutant PTP␣ to activate Src in vitro was measured after immunopurifying the HA-tagged phosphatase from unsynchronized or mitotic overexpresser cells. Equal amounts of solubilized phosphatase were incubated in phosphatase buffer with chicken WT Src that had been immunoprecipitated from unsynchronized NIH3T3-derived Src overexpresser cells. After washing away PTP␣, the specific activity of the treated Src was measured by incubating it with [␥-32 P]ATP and acid-denatured enolase (substrate) and measuring the amount of transferred 32 P by autoradiography ( Fig. 2A, panel  a). Because Tyr(P) 527 is the only detectable phosphotyrosine in Src from these overexpresser cells (31), anti-Tyr(P) immunoblotting was used to assay the amount of Tyr 527 phosphorylation ( Fig. 2A, panel b). As previously shown (24), PTP␣ from mitotic cells dephosphorylated Src and increased its kinase activity more than PTP␣ from unsynchronized cells ( Fig. 2A,  lanes 3 and 4). We now found that Ser-to-Ala mutation of either Ser 180 or Ser 204 or both not only abrogated the mitotic increase in activity, but largely eliminated the ability of PTP␣ from both unsynchronized and mitotic cells to dephosphorylate and activate Src at all (Fig. 2A, lanes 7-12). The very low residual activity of the Ser-to-Ala mutants was similar to that of the Y789F mutant ( Fig. 2A, lanes 5 and 6).
Additional experiments were conducted in which PTP␣ was dephosphorylated by PP2A prior to incubation with the Src substrate. Phosphatase-treated WT PTP␣ from both unsynchronized and mitotic cells had the same very low Src-activating ability as untreated PTP␣(S180A/S204A) (Fig. 2B, panel a,  lanes 3-6). Moreover, PP2A treatment did not affect PTP␣(S180A/S204A) (Fig. 2B, lanes 5-8), suggesting that its effect on WT PTP␣ was mediated via dephosphorylation of Ser 180 and Ser 204 .
To assess the Src-directed activity of WT and mutant PTP␣ in vivo, Src was immunoprecipitated from unsynchronized and mitotic non-overexpresser cells (control) and PTP␣ overexpresser cells, and Src phosphorylation and its ability to phosphorylate enolase were measured (Fig. 3, a and b). All of the overexpresser cell lines expressed approximately equal amounts of Src and transgene PTP␣ (Fig. 3, c and d). As previously shown (24), overexpression of WT PTP␣ decreased Src tyrosine phosphorylation, increased interphase Src activity, and enhanced the mitotic increase in its activity (Fig. 3, 3 and 4) or mutant (lanes 5-12) PTP␣-HA that had been immunopurified from unsynchronized (U) or mitotic (M) PTP␣ overexpresser cells using anti-HA antibody or with mock-immunopurified protein from control cells that did not express any HA-tagged protein (lanes 1 and 2). The partially dephosphorylated Src immunoprecipitates were washed to remove PTP␣ and then incubated with enolase and [␥- 32 1 and 2) or from cells overexpressing WT (lanes 3 and 4) or mutant (lanes 5-10) PTP␣ as indicated. Each immunoprecipitate was divided into aliquots that were used in a kinase assay with [␥-32 P]ATP and acid-denatured enolase, followed by electrophoresis and autoradiography of the reaction products (a); immunoblotted with anti-Tyr(P) (Anti-pTyr) antibody (b); immunoblotted with anti-Src antibody (c); or immunoblotted with anti-PTP␣ antibody (d). SDS-PAGE was performed on 9% gels. The positions of molecular mass markers are indicated in kilodaltons.

FIG. 3. Effect of WT and mutant PTP␣ overexpression on Src in vivo tyrosine phosphorylation and kinase activity. Endogenous (Endog) Src was immunoprecipitated from unsynchronized (U) or mitotic (M) non-overexpresser cells (lanes
compare lanes 3 and 4 with lanes 1 and 2). In contrast, we now found that overexpression of the mutants with S180A and/or S204A substitutions had no effect on Src tyrosine phosphorylation or activity (Fig. 3, lanes 5-10). In summary, the in vitro and in vivo results consistently imply that phosphorylation of PTP␣ at both Ser 180 and Ser 204 is required for it to be able to dephosphorylate and activate Src.
Mutation of Ser 180 or Ser 204 , but Not of Ser 202 , Blocks Neoplastic Transformation by PTP␣-The ability of the WT and mutant PTP␣ overexpresser cells to grow without anchorage was assayed by suspending them in semisolid medium containing 0.3% soft agarose without doxycycline (Fig. 4). The expression levels in the WT and mutant overexpresser cells were the same, except for the S180A mutant, which was expressed at an ϳ50% higher level. Overexpression of WT PTP␣, but not of either the coordinate or separate Ser-to-Ala mutants, induced anchorage-independent growth. A mutant that contained a S202A mutation transformed like WT PTP␣. Similar results were obtained with cells overexpressing WT and mutant PTP␣-HA proteins (data not shown).
Effect of Ser 180 and Ser 204 Mutations on PTP␣ Binding to Src and Grb2-The association of WT and mutant PTP␣ with Src in vivo was examined by co-immunoprecipitation experiments. PTP␣ overexpresser cells were lysed with a Nonidet P-40 buffer, and anti-Src immunoprecipitates were immunoblotted with anti-PTP␣ (Fig. 5a) or anti-Src (Fig. 5b) antibody. As previously reported (24), the association between WT PTP␣ and Src increased ϳ3-fold in mitotic cells (Fig. 5a, compare  lanes 3 and 4). In contrast, PTP␣ with either the S180A and/or S204A mutation did not detectably bind Src in either unsynchronized or mitotic cells (Fig. 5a, lanes 5-10). (The lower amount of endogenous PTP␣ in the control cells was not detectable in these experiments and so did not influence the observed results.) Analogous co-immunoprecipitation experiments were performed to examine PTP␣ binding to Grb2. Anti-Grb2 and anti-PTP␣ immunoprecipitates were immunoblotted with anti-PTP␣ and anti-Grb2 antibodies, respectively (Fig. 6, A and B). Both types of experiments gave consistent results. As previously reported (13,24), the WT PTP␣-Grb2 association observed in unsynchronized non-overexpresser and overexpresser cells decreased ϳ4-fold in mitotic cells. (As previously noted, Grb2 bound a larger fraction of endogenous WT PTP␣ (i.e. in the Neo control cells) than overexpressed WT PTP␣. This is because the level of Tyr 789 phosphorylation in overexpressed PTP␣ is ϳ2.5 times lower relative to endogenous PTP␣, possibly because the Tyr 789 kinase is saturated (13).) In contrast, we now found that the Ser-to-Ala PTP␣ mutants bound Grb2 to the same high extent in both unsynchronized and mitotic cells. The control immunoblots (Fig. 6, A, lanes 1, 2, and  5-8; and B, panels a, c, and d) showed that there were similar amounts of Grb2 and PTP␣ in the cell lysates and immunoprecipitates from the unsynchronized and mitotic cells. We conclude that both Ser 180 and Ser 204 are required for PTP␣ binding to Src in vivo and for the mitotic reduction in PTP␣-Grb2 association.
Effect of Ser 180 and Ser 204 Mutations and PP2A Treatment on PTP␣ Binding to the Isolated Src and Grb2 SH2 Domains-To determine whether the changes in PTP␣-Grb2 association could be explained by changes in the affinity of the Grb2 or Src SH2 domain alone, we measured the abilities of fusion proteins containing GST and either Grb2 or Src SH2 domains to affinity-precipitate WT PTP␣-HA and PTP␣(S180A/S204A)-HA from overexpresser cell lysates (Fig.  7A). As previously shown (13), the Grb2 SH2 domain immunoprecipitated about three times more interphase PTP␣ than the Src SH2 domain (Fig. 7A, compare lanes 5 and 7). Also as previously reported (24), the Grb2 SH2 domain bound ϳ2fold less mitotic PTP␣ than interphase PTP␣ (Fig. 7A, compare lanes 5 and 6), but the affinity of the Src SH2 domain for mitotic PTP␣ was the same as or possibly slightly higher than its affinity for interphase PTP␣ (compare lanes 7 and 8). We now found that the Ser-to-Ala mutations slightly increased (ϳ25%) the ratio between Grb2 and Src SH2 domain binding to interphase PTP␣ and completely eliminated the mitotic changes in binding affinities; 3.7 Ϯ 0.5 times more PTP␣(S180A/S204A) bound to the Grb2 SH2 domain than to the Src SH2 domain (Fig. 7A, lanes 13-16).
To exclude the possibility that other proteins in the cell lysates affected PTP␣ binding to the GST-SH2 domain fusion proteins, similar experiments were performed with immunopurified PTP␣-HA (Fig. 7B). (The immunopurification procedure involved a pH 2.5 elution and subsequent neutralization that removed all co-associated Grb2 (data not shown) and presum- ably any other noncovalently associated proteins.) Similar results were obtained (Fig. 7B, lanes 5, 6, 9, and 10). We also examined the effect of serine dephosphorylation on the binding of immunopurified PTP␣-HA by incubating it with PP2A before the affinity precipitations. Dephosphorylation affected WT PTP␣-HA binding in the same manner as the S180A and S204A mutations: it eliminated the mitotic decrease in the binding of PTP␣ to the Grb2 SH2 domain and decreased the binding of interphase and mitotic PTP␣ to the Src SH2 domain by 35-45%. We conclude that mitotic phosphorylations at Ser 180 and Ser 204 are required for the 2-3-fold mitotic downregulation of the affinity of the Grb2 SH2 domain for PTP␣. In contrast, the phosphorylations slightly increase binding to the Src SH2 domain. DISCUSSION We have previously shown that, during mitosis, the catalytic activity of PTP␣ is enhanced and that its inhibitory binding to Grb2, which specifically blocks Src dephosphorylation, is decreased (24). We have now shown that S180A and/or S204A mutation blocks these effects and the resultant mitotic activation of Src by PTP␣. This occurs without change in the phosphorylation of PTP␣ at Tyr 789 , which is required for phosphotyrosine displacement and Src dephosphorylation.
Surprisingly, these mutations also prevent Src-PTP␣ co-association during interphase and block most or all dephosphorylation and activation of Src both in vitro and in vivo. This almost certainly explains the inability of any of the mutants to induce anchorage-independent growth, even when expressed at high (ϳ20 times endogenous) levels. The mutations do not prevent dephosphorylation of MBP, implying that Ser 180 and Ser 204 , like Tyr 789 , can regulate PTP␣ substrate specificity.
The fact that either Ser 180 /Ser 204 mutation or PP2A treatment removes the mitotic electrophoretic mobility retardation of PTP␣ indicates that these residues have been hyperphosphorylated. Moreover, the similarities between the mutation-and PP2A-induced effects on catalytic activity and binding indicate that the mutations act functionally by preventing this hyperphosphorylation. Protein kinase C phosphorylates Ser 180 and Ser 204 following 12-O-tetradecanoylphorbol-13-acetate stimulation (25, 26), and it may phosphorylate them at or shortly before mitosis. This could account for the observed changes in catalytic activity: 12-O-tetradecanoylphorbol-13-acetate-stimulated protein kinase C-mediated phosphorylation at these sites decreases the PTP␣ K m for MBP from 12 to 5 M without significant change in V max (25). At the concentration of tyrosine-phosphorylated MBP in our assays (ϳ4 M), such a 2.4-fold reduction in K m would cause an ϳ1.8-fold increase in PTP activity, consistent with our measurements.
The protein kinase C isoform that is most likely to be involved is protein kinase C␦, which co-associates with PTP␣ (and phosphatidylinositol 3-kinase) following treatment of A431 cells with a somatostatin analog (27). This results in activation of PTP␣ and Src, probably initiated by protein kinase C␦-mediated phosphorylation of Ser 180 and Ser 204 (27,32). Protein kinase C has been implicated in both positive and negative control of the G 2 /M transition, with the relevant However, the participation of other kinases is not excluded. For example, the sequence surrounding Ser 204 also matches the protein kinase A phosphorylation consensus sequence (35). Whatever the kinase, because dephosphorylation of Src by PTP␣ requires phosphorylation at both Ser 180 and Ser 204 , this activity will be proportional to the square of the serine phosphorylation stoichiometry (assuming that the phosphorylations are independent events). This non-linearity will enhance the ability of the upstream serine kinase to control PTP␣ and hence Src in an "on-off " manner.
The fact that mutation of either Ser 180 or Ser 204 prevents the mitotic decrease in the binding of PTP␣ and Grb2 in vivo (Fig.  6) suggests that coordinate phosphorylation at these sites during mitosis reduces their binding affinity. As described in the Introduction, very little Grb2-unbound, Tyr 789 -phosphorylated PTP␣ is available to bind and act on Src during interphase. Thus, it is likely that the inability of PTP␣(S180A/S204A) to bind Src in vivo results, at least in part, from increased competition from Grb2. The fact that the mutations affect PTP␣ during interphase (as well as during mitosis) is consistent with the observation that Ser 180 and Ser 204 are phosphorylated to some extent in unsynchronized cells (26).
The mitosis-and mutation-induced changes in PTP␣-Grb2 binding in vivo correlate perfectly with and may result from the corresponding changes observed in the affinity between PTP␣ and the Grb2 SH2 domain in vitro (Fig. 7). The in vitro binding experiments were performed using recombinant GST-SH2 domain fusion proteins, excluding the possibility that a cell cycledependent modification of Grb2 or altered binding to a Grb2 SH3 domain is required. The most economical hypothesis, supported by both the mutagenesis and PP2A dephosphorylation experiments, is that phosphorylation of Ser 180 and Ser 204 decreases PTP␣-Grb2 SH2 domain affinity by 3-4-fold and increases PTP␣-Src SH2 domain affinity by 35-45%.
Both the Grb2 and Src SH2 domains bind to Tyr(P) 789 , which is the only phosphotyrosine in PTP␣ (11,13); so it is surprising that their binding affinities can be differentially regulated. As far as we are aware, this has no precedent. The binding affinity changes were observed with immunopurified PTP␣ that had passed through a pH 2.5 denaturation step that removed all Grb2 (and probably any other co-associated proteins), and PTP␣ dimerization was not detected in the cell lysates used (data not shown). Therefore, we believe that the changes reflect effects of the Ser 180 and Ser 204 phosphorylations on isolated monomeric PTP␣.
The mechanism of differential regulation may be related to the unique mode by which the Grb2 domain binds with high affinity to Tyr(P)-containing peptides: peptides that match the Grb2 SH2 domain binding consensus sequence form a ␤-turn when bound to the SH2 domain (36,37). This is probably the conformation of the Tyr(P) 789 region when bound to the Grb2 SH2 domain with high affinity. In contrast, Tyr(P)-containing peptides bind to other SH2 domains in an extended conformation (38,39), so the Tyr(P) 789 region is probably extended when it binds the Src SH2 domain. Therefore, it is possible that phosphorylation of Ser 180 and Ser 204 could reduce the high affinity binding to the Grb2 SH2 domain without decreasing the lower affinity binding to the Src SH2 domain if it interfered with the formation of the ␤-turn. If the phosphorylations also stabilized an extended conformation of the C-terminal region, they could simultaneously increase the affinity of Tyr(P) 789 binding to the Src SH2 domain.
A model of this sort is shown in Fig. 8. In this hypothesis, phosphorylations at Ser 180 and Ser 204 promote an intramolecular association between the PTP␣ C-terminal and membraneproximal regions that prevents the ␤-turn and stabilizes an extended conformation of the C-terminal region without occluding Tyr(P) 789 . For example, basic residues surrounding Tyr(P) 789 (e.g. Lys 777 and Lys 793 ) might interact with phosphorylated Ser 180 and Ser 204 so as to "stretch" the peptide out along the surface of PTP␣. To speculate further, the intramolecular association might also increase the accessibility (or modify the conformation) of the D1 catalytic domain, which lies between these two regions, so as to decrease its K m . Modulation of intramolecular association by changes in Ser 180 and Ser 204 phosphorylation could thereby also account for the observed changes in nonspecific catalytic activity.
Our results do not exclude the possibility that phosphorylation of Ser 180 and Ser 204 is required but not sufficient for activation. Of particular interest is Ser 787 , the only serine in the C-terminal region downstream from the D2 catalytic domain. It lies in a (weak) phosphorylation consensus site for casein kinase II (40) and, because of its proximity to Tyr 789 , FIG. 8. An intramolecular association hypothesis. A, when either Ser 180 or Ser 204 is dephosphorylated, the C terminus of PTP␣ is free to adopt a ␤-hairpin conformation and bind the Grb2 SH2 domain. B, when both Ser 180 and Ser 204 are phosphorylated, an intramolecular interaction between the C terminus and the membrane-proximal region stabilizes an extended conformation of the C terminus that reduces its affinity for the Grb2 SH2 domain while facilitating binding by the Src SH2 domain. In the variant of the hypothesis shown here, this association also relieves partial occlusion of the D1 catalytic site by the D2 domain, thereby increasing catalytic activity. pY, Tyr(P); pS, Ser(P). might directly affect its interaction with SH2 domains if it were also hyperphosphorylated during mitosis.
In any case, at least one puzzle remains: although we expect altered binding and increased competition from Grb2 to reduce the ability of the Ser-to-Ala PTP␣ mutants to dephosphorylate Src in vivo, it does not explain their inability to dephosphorylate Src in vitro under conditions in which Grb2 (and probably any other co-associating proteins) was removed. Even though its binding affinity is slightly reduced, PTP␣(S180A/S204A) still binds the Src SH2 domain (Fig. 7), so it is not evident why phosphotyrosine displacement and dephosphorylation of Src should be blocked. Because Tyr(P) 789 must compete with Tyr(P) 527 for binding to the Src SH2 domain, it is possible that even a fairly small change in Src SH2 domain-Tyr(P) 789 affinity can perturb a delicate balance. However, other mechanisms may also be involved.