Expression, Purification, and Characterization of SH2-containing Protein Tyrosine Phosphatase, SH-PTP2*

Ahuman protein tyrosine phosphatase containing two src homology 2 (SH2) domains (SH-PTP2) was expressed in Escherichia coli under T7 promoter control and puri- fied to near homogeneity. The purified protein, with molecular mass of 68 kDa on SDS-polyacrylamide gel elec- trophoresis, was identified as SH-PTP2 by its protein tyrosine phosphatase activity and N-terminal amino acid sequence analysis. Its protein tyrosine phosphatase activity was sensitive to pH and salt concentration. Whereas its optimum pH for the low molecular weight substrate para-nitrophenyl phosphate is 5.6, the pH op-tima for peptide substrates were shifted toward neutral. With the artificial protein substrate reduced, carboxy-amidomethylated, and maleylated lysozyme, it displays 2000-fold lower K, (1.7 px) and 2.4-fold higher kcet (0.11 s-l) than with para-nitrophenyl phosphate. Among the phosphopeptides from autophosphorylation sites of receptors for epidermal growth factor and platelet-de- rived growth factor, SH-PTP2 displayed high activity toward phosphopeptides corresponding ligated NdeI-SaZI-linearized PET-SH-PTP1 plasmid PET-SHPTP2. The sequence from the Shine-Dalgano sequence to 15th amino acid from the N terminus was confirmed by dideoxy sequencing using the T7 promoter primer, 5"TAATACGACT- CACTATAGGG-3' (20-mer). of reaction mixture was transferred to a suspension of activated charcoal. From the counts in the supernatant, dephosphorylation velocities were calculated. Other Methods-The concentration of protein was determined by Bradford assay (Bio-Rad) using BSA as standard (48). SDS-PAGE was carried out as described by Laemmli (49). Gel filtration was carried out with Superose 12 (10/30, Pharmacia) equilibrated with 25 mM HEPES, pH 7.2, containing 200 mM NaCl, at a flow rate of 0.5 mumin using aldolase (M, = 158,000), BSA (M, = 67,000), and ovalbumin (M, = 43,000) as standards. N-terminal sequencing was performed by the Edman method (50) using an automated gas-phase sequenator (Applied Biosystems).

membrane protein tyrosine phosphatases, we have cloned , also known as PTPlC (151, HCP (16), SHP (17), and PTPN6 (18) and , also known as PTPlD (21), and PTP2C (221, both of which contain two SH2 domains upstream from the conserved catalytic domain. Mouse Syp is reported as a homologue of SH-PTP2 based on the high similarity of the cDNA sequence (23). Rat PTPLl (24) may be also a homologue of SH-PTP2, based on the high similarity in reported partial amino acid sequences of PTPLl with SH-

PTP2.
SH2l domains are also found in several other types of signaling proteins, such as src family protein tyrosine kinases, GTPase-activating protein, phospholipase C-y, and p85, the regulatory subunit of phosphatidylinositol 3-kinase (25). SH2 domains bind to tyrosine-phosphorylated sequences in proteins and peptides, thereby facilitating inter-and intramolecular protein-protein interactions, including enzyme-substrate interactions (26). Recently, phosphotyrosine-independent binding to SH2 domains has also been reported (27). The finding of protein tyrosine phosphatases with SH2 domains that would, presumably, target these enzymes to specific phosphotyrosine-containing protein substrates is of distinct physiological interest.
Although SH-PTP1 is predominantly expressed in hematopoietic cells (14,16,17), SH-PTP2 is ubiquitously expressed (19,20,22). Based on its sequence similarity and comparable expression pattern, SH-PTP2 may be the homologue of the Drosophila corkscrew gene product (Csw) (19,281. Genetic epistasis experiments indicate that Csw functions in the terminal class signal transduction pathway in concert with the Drosophila c-Raf homologue (l(1) polehill gene product (29) or D-Raf), to positively transduce signals generated by the torso receptor protein tyrosine kinase (30, 311, a PDGF receptor (PDGFR) homologue.
We and others (21,23,32) have shown that SH-PTP2 is tyrosyl-phosphorylated in vivo upon activation of the EGF receptor (EGFR) or the PDGFR, although this has not been shown to be a direct effect of the receptor kinase. Moreover, we have found that SH-PTP2 is directly bound to EGFR and PDGFR via its N-terminal SH2 domain following ligand activation (32). Since the EGFR (33)(34)(35) and the PDGFR ( 3 6 4 0 ) are protein tyrosine kinases and are autophosphorylated within the cytoplasmic domain, it seems likely that SH2 domains in SH-PTP2 will play a crucial role in binding to the EGFR and PDGFR via their autophosphorylation sites and for further signal transduction. SH-PTP2 binds to phosphorylated The abbreviations used are: SH2, src homology 2; EGF, epidermal growth factor; EGFR, EGF receptor; PDGF, platelet-derived growth factor; PDGFR, PDGF receptor; p-NPP, para-nitrophenyl phosphate; RCM-lysozyme, reduced, carboxyamidomethylated, and maleylated lyfonate; DTT, dithiothreitol; BSA, bovine serum albumin. sozyme; pY, phosphotyrosyl residue MES, 2-(N-morpholino)ethanesul- In this paper, we describe the expression, purification, and enzymatic properties of recombinant SH-PTP2 derived from Escherichia coli, including its interactions with phosphopeptides corresponding to autophosphorylation sites of EGFR and PDGFR as protein tyrosine phosphatase substrates and effectors.
EXPERIMENTAL PROCEDURES Materials-Restriction endonucleases NdeI and SalI were purchased from New England Biolabs and ThaI from Life Technologies, 1nc.p-NPP was from Sigma. Reduced, carboxyamidomethylated, and maleylated (RCM) lysozyme was from Life Technologies, Inc. v-Ab1 was from Oncogene Science. The expression vector PET-SHPTP1 (41) was kindly donated by D. Pei (Harvard Medical School). Oligonucleotide adapters Medical School). Phosphopeptide EGFR pY1173 was kindly donated by and sequencing primers were synthesized by A. Nussbaum (Harvard H. Cho (Harvard Medical School).
Expression and Purification of SH-PTPZ-E. coli strain BL21 (DE3) transformed with plasmid PET-SHPTP2 was grown in 4 liters of LB medium containing 50 pg/ml ampicillin at 37 "C to a n absorbance at 595 nm of 0.8 and induced for 3 h a t 30 "C with 0.4 mM isopropyl-l-thio-a-D-galactopyranoside. Cells were harvested by centrifugation and resuspended in 100 ml of 10 mM Tris-HC1, pH 7.8, containing 1 mM EDTA and 10 mM 2-mercaptoethanol (buffer A) supplemented with protease inhibitors (0.5 mM ortho-phenanthroline, 0.64 mM benzamidine, 0.29 mM phenylmethylsulfonyl fluoride, 20 pg/ml soybean trypsin inhibitor, 20 pg/ml aprotinin, 20 pg/ml leupeptin, and 20 pg/ml pepstatin). The cells were disrupted by French Press, and the crude lysate was centrifuged at 15,000 rpm for 15 min in a Sorvall SS-34 rotor. The supernatant was loaded onto a Q-Sepharose Fast Flow (Sigma) column (11.2 x 2.5 cm) equilibrated with buffer A. The column was washed with 200 ml of buffer A, and 0-500 mM NaCl concentration gradient was developed in 400 ml of buffer A at 0.5 mumin. Activity, recovered in the flow-through and very early part of gradient elution, was precipitated by 60% saturated ammonium sulfate, dissolved in buffer A containing 20% saturated ammonium sulfate, and loaded onto a phenyl-Sepharose (Sigma) column (13.2 x 2.5 cm) equilibrated with buffer A containing 20% saturated ammonium sulfate. The column was washed with the same buffer, and activity was eluted using an ammonium sulfate concentration gradient, 20 to 0%, in 400 ml of buffer A a t 0.5 mumin. The active fractions, eluted at approximately 8% saturated ammonium sulfate, were dialyzed against 10 mM MES, pH 5.7, containing 10 mM 2-mercaptoethanol (buffer B), and divided into four aliquots of equal volume. Each aliquot was loaded onto Mono S HR 10/10 (Pharmacia LKB Biotechnology Inc.) column equilibrated with buffer B. The column was developed with a gradient of 0-250 mM NaCl in 500 ml of buffer B a t 2.5 ml/min. Pooled fractions, eluting from 125 to 150 mM NaC1, were concentrated using a Centriprep-10 (Amicon) and stored a t -80 "C in the presence of 33% (v/v) glycerol.
Assay for Protein Qrosine Phosphatase Actiuity-Withp-NPP a s substrate, typically 10 mMp-NPP was incubated with 87 pg/ml SH-PTP2 a t 24 "C for 130 min in 50 pl of 50 mM 3,3-dimethylglutarate, pH 5.6, containing 50 mM NaCl, 10 mM DTT, and 2 mM EDTA. The reaction was quenched with 950 pl of 1 M NaOH, and the absorbance of p-nitrophenolate at 405 nm was measured. The amount ofp-nitrophenol released were calculated by comparison with a standard curve obtained with p-nitrophenolate (Sigma). To assay the dephosphorylation of phosphopeptides, the release of Pi was measured by malachite green assay (4244). Typically, 500 PM phosphopeptide was incubated with 87 pg/ml SH-PTP2 a t 20 "C for 12 min in 50 pl of 50 mM HEPES, pH 7.1, containing 150 mM NaCI, 10 mM DTT, and 2 mM EDTA. The reaction was quenched with 950 pl of malachite green reagent, and the absorbance a t 650 nm was measured. The amount of released Pi was calculated with a standard curve. Phosphopeptides were synthesized as described in Piccione et al. (45) using the methodology of Kitas et al. (46). Sequence of phosphopeptides used here were as follows: EGFR pY992, DADEpYLIPQQGFF; EGFR pY1068, WEpYINQSVPK, EGFR pY1086, NVPpYHNGPLNP; EGFR pY1148, NPDpYQQDFFPK EGFR pY1173, TAENAEpYLRVA, PDGFR pY740, DGGpYMDMSKDE; PDGFR pY751, SVDpYVPMLDMK PDGFR pY771, S S N p W Y D N Y ; PDGFR pY1009, SVLpYTAVQPNE; PDGFR pY1021, DNDpYIIPLPDPK, with pY indicating the phosphorylated tyrosine.
To assay the dephosphorylation of a protein substrate, phosphorylated RCM-lysozyme was prepared essentially by the method of Tonks et al. (47), but using v-Ab1 as the kinase. Typical specific activity obtained was 13 pCi/mol. Assay conditions were essentially the same as those of Tonks (47). The indicated concentration of phosphorylated RCM-lysozyme was incubated with 320 ng/ml SH-PTP2 at 30 "C for 5 min in 60 pl of reaction buffer consisting of 25 mM HEPES, pH 7.2, containing 1 mg/ml bovine serum albumin (BSA), 5 l l l~ EDTA, and 10 mM DTT. At 1 and 5 min, 25 pl of reaction mixture was transferred to a suspension of activated charcoal. From the counts in the supernatant, dephosphorylation velocities were calculated.

RESULTS
Expression and Purification of SH-P!l'P2-Previously we observed that SH-PTP1, a protein highly similar to SH-PTP2, was highly expressed in E. coli and that when expressed in bacterial cells, SH-PTP1 accumulated partially as a soluble protein (41). Therefore, we used the same expression vector with the coding region of SH-PTP2. The expression level of SH-PTP2 was low, approximately 1% of total E. coli cell protein compared with about 10% for SH-PTP1. However, most of the activity was recovered in the soluble fraction (data not shown).
Soluble bacterially expressed SH-PTP2 was purified according to the scheme summarized in Table I. During the purification,p-NPP was used as substrate. E. coli alkaline phosphatase also reacts with p-NPP. During the first three steps, SH-PTP2 is likely contaminated with alkaline phosphatase, suggesting that the yield for the last three steps is higher than the values in Table I. Using this scheme, SH-PTP2 was purified to greater than 90% purity based on SDS-PAGE analysis (Fig. 1). The molecular mass of the purified protein was 68 kDa, consistent with the 593-amino acid sequence and similar to the in vivo molecular mass of 68-70 kDa (21,23,32). The purified enzyme behaved as a monomer on gel filtration (data not shown). The N-terminal sequence of the purified protein was analyzed up to the 10th residue and matched the predicted sequence from residue 2 on, indicating processing of the initiating N-terminal methionine.
Biochemical Characterization of SH-PTP2 Phosphatase Activity toward p-NPP-We elucidated the optimum pH, salt concentration, and temperature for SH-PTP2 activity initially USing the low molecular weight substratep-NPP. As shown in Fig.  2, like SH-PTP1 (41) and other protein tyrosine phosphatases (51), SH-PTP2 showed a n acidic pH optimum with this substrate, pH 5.6. As determined by the NaCl concentration profile for SH-PTP2, the optimal salt concentration for SH-PTP2 activity towardp-NPP was 50 mM a t pH 5.6 with only 9% activity
Properties of SH-PTP2 toward Phosphopeptide /?om the EGFR and the PDGFR Autophosphorylation Sites-Given the in vivo evidence for specific binding of SH-PTP2 to EGFR and PDGFR (32), we have evaluated the properties of autophosphorylation site peptides from the cytoplasmic domains of each of these transmembrane growth factor receptors as substrates of and/or effectors for SH-PTP2. An initial screen utilized the synthetic 11-13-amino acid phosphopeptides corresponding to each known autophosphorylation site (33)(34)(35)(36)(37)(38)(39)(40).
As indicated in Fig. 4A, on incubation at pH 5.6 (the pH optimum determined for p-NPP (see Fig. 2)), the EGFR pY992 was dramatically better as a substrate than the other EGFR autophosphorylation site peptides. When the EGFR pY992 and EGFR pY1173 phosphopeptides were compared ("good" and "bad" substrates, respectively), EGFR pY1173 showed a pH optimum at pH 6.1 (Fig.2), but the EGFR pY992 substrate showed a pH optimum in the physiological pH range (Fig. 2). Thus, subsequent characterization was done under physiological conditions. At neutral pH, the EGFR phosphopeptides show a similar profile to that at acidic pH (Fig. 4, B versus A ) with EGFR pY992 showing >lO-fold preferential substrate activity for SH-PTPB.
In comparison with EGFR peptides, the distinction between the five phosphopeptides from the PDGFR at pH 5.6 was less (Fig. 4A); PDGFR pY1021 was chosen as was PDGFR pY1009 for subsequent kinetic evaluation. At physiological pH, PDGFR pY1009 is now preferred about 2-fold over PDGFR pY1021 and 4-10-fold over the other sites (Fig. 4B).   Michaelis-Menten hyperbolic saturation behavior were detected. Fig. 5 shows a Lineweaver-Burk plot of PDGFR pY1009 which reproducibly shows curvature. A simple linear analysis from data at low substrate concentration would yield a y axis crossing in the negative region, which precludes calculation of the K , and kcat. However, for data at higher substrate concentration, the slope approaches linear behavior, and an extrapolated line gave a V, , , estimate of 20 pdmin (inset of Fig. 5). It may be that this peptide shows both allosteric and substrate effects.

DISCUSSION
In this work, the full-length human SH2 domain-containing protein tyrosine phosphatase, SH-PTP2, was expressed in E. coli and highly purified. Basic enzymatic properties have been assessed with a low molecular weight substrate p-NPP and subsequently with 11-13 residue phosphopeptides and tyrosine-phosphorylated RCM-lysozyme.
Protein tyrosine phosphatase activity of SH-PTP2 toward p-NPP shows an acidic pH optimum and sensitivity to ionic strength, similar to SH-PTP1 (41). From the Arrhenius plot (Fig. 3), SH-PTP2 was stable for 130 min at pH 5.6 below 24 "C, but was labile above 24 "C. Although assays with RCM-lysozyme were done at 30 "C for 5 min, our data were within linear regions of Pi release versus time, indicating no inactivation during the reaction.
In Table 11, K , and kcat values for p-NPP are compared with those reported for other protein tyrosine phosphatases (22,41,51,(53)(54)(55)(56). As a p-nitrophenylphosphatase, SH-PTP2 is dramatically slow, even under optimum pH and ionic strength conditions. It has 30,000-fold lower kcat than the Yersinia protein tyrosine phosphatase Yop (51). The kcat of SH-PTP2 as a p-nitrophenylphosphatase is lower than that of VH1 (51) and slightly higher than that of cdc25 (55). At 10 m M p-NPP, SH-PTP2 has about 2% the catalytic activity of the highly similar, E. coli-expressed, SH-PTP1 (41 and data not shown). To determine that SH-PTP2 expressed was not largely inactive, we compared the activity of SH-PTP2 toward phosphotyrosyl-RCM-lysozyme with that of SH-PTP1 (PTPlC) (56) and with the value just reported for SH2 domain-deleted SH-PTP2 mutant (ASH2-PTP2C) (22) (Table 111). Kinetic data for SH-PTPZ are almost the same as those of SH-PTP1 (PTPlC) (561, and the kcat of SH-PTP2 is about one-third the value of the SH2 domain-deleted SH-PTPZ mutant (22). We also compared the activity of SH-PTP2 toward the phosphotyrosyl peptides EGFR pY992 and EGFR pY1173 with SH-PTP1 expressed in E. coli (41). SH-PTP2 is only 5-8-fold slower than S H -m P l under the same assay conditions (data not shown). These comparisons are consistent with the view that the purified SH-FTP2 is correctly folded full-length enzyme. It may be worthwhile to express full-length SH-PTP2 in a eukaryotic overproduction system (e.g. baculovirus) to compare catalytic efficiency.
Since SH-PTP2 has SH2 domains and protein tyrosine phosphatase domain, it is possible that phosphotyrosyl proteins serve as both substrates and SHZ domain targets. Using phosphotyrosyl peptide substrates from the EGFR (33)(34)(35) and PDGFR (36)(37)(38)(39)(40) cytoplasmic domains, SH-PTP2 shows high activity for three of the 10 autophosphorylation site peptides (Fig. 4). For PDGFR pY1009 which has been examined carefully, the anomalous dependence of velocity on pY peptide concentration may reflect both allosteric and active site binding.
Songyang et al. (57)  from a synthetic peptide library as an enriched ligand for the N-terminal SH2 domain of SH-PTP2. PDGFR pY1009 has the sequence pYTAV, approximating the enriched sequence. Moreover, we recently found that pY1009 is the binding site in PDGFR for glutathione S-transferase fusion proteins with SH2 domains derived from SH-PTP2.' Initial binding studies of PDGFR pY1009 to the isolated SH2 domains of SH-PTP2 show binding with differential affinity3 These results support the assumption that either one or both of the SH2 domains is the allosteric site for PDGFR pY1009.
For further characterization of the allosteric effect of PDGFR pY1009 phosphotyrosyl substrate on SH-PTP2, the assay system we present in this work is complicated because PDGFR pY1009 acts as substrate and effector. Therefore, we are trying to establish a system to separate substrate and effector. Recently we find that addition of PDGFR pY1009, but not the other nine phosphotyrosine peptides of Fig. 4, substantially stimulates enzymatic activity toward dephosphorylation of phosphotyrosyl-RCM-lysozyme, which by itself has no allosteric effect on enzymatic activity, at least in the range of concentration used in this work.' These results suggest that occupation of the SH2 domains may serve to up-regulate phosphatase activity.
SH-PTP2 demonstrates substantial protein tyrosine phosphatase activity toward peptides EGFR pY992 and PDGFR pY1021 in addition to PDGFR pY1009 (Fig. 4B). In contrast to PDGFR pY1009, EGFR pY992 and PDGFR pY1021 have some common sequence properties, such as more than 2 acidic amino acid residues N-terminal to the phosphotyrosine, aliphatic amino acid residues at both +1 and +2 position, and Pro at +3 position. By these properties, EGFR pY992 and PDGFR pY1021 are distinct from other phosphopeptides derived from autophosphorylation sites of the EGFR receptor and PDGFR, suggesting that these peptides might provide good phosphotyrosyl substrates for SH-PTPZ.
Zhang et al. (58) have determined the specificity of the catalytic domains of protein tyrosine phosphatase from human and Yersinia, toward phosphotyrosyl peptides and acidic amino acid residues at positions -1, -2, and -4 and Pro at position +3 are important for high activity. If the rules suggested by Zhang et al. (58) are applicable to SH-PTP2, EGFR pY992 corresponds to the best substrate derived from EGFR autophosphorylation sites, and PDGFR pY1021 also matches the sequence determinants for a good substrate. Interestingly, both of these two sites are also PLC-7 binding site (34, 59); however, the biological significance of this finding is not clear.
Limited tryptic cleavage of the 68-kDa intact SH-PTP2 to yield a 65-kDa fragment also increases protein tyrosine phosphatase activity by f~u r -f o l d .~ Similarly, after partial tryptic digestion resulting in cleavage of the C-terminal 5-kDa fragment, SH-PTP1 (PTPlC) enzymatic activity increased about 20-fold (56). We have not confirmed the cleavage site yet for SH-PTP2, but if tryptic digestion of SH-PTP2 also removes the C-terminal region, then this would be an indication that the C-terminal tail of SH-PTP2, as in SH-PTP1, also has negative autoregulatory effect.
So far we have noted the stimulatory or inhibitory effects of phosphopeptides from receptor kinases and SH2 domains and the C-terminal regions of SH-PTPs. It is important, we believe, S. Sugimoto and C. T. Walsh, unpublished results. to clarify the relationship of these effects. "0 resolve these issues, we plan to compare the properties of SH2 and C-terminal domain-deleted or -substituted mutants and examine the contribution of each domain to both binding and catalytic activities using phosphorylated and nonphosphorylated peptides or proteins.
During the preparation of this manuscript, two papers have appeared, one describing the in vivo characterization of SH-PTP2 in a transient expression system (21) and another characterizing the murine homologue of SH-PTP2 (23). In the former paper, SH-PTP2 (named PTPlD by these authors) was co-transfected into 293 cells with receptor chimeras composed of the extracellular domain of the EGFR and cytoplasmic domain of the PDGFR. Immunoprecipitated SH-PTP2 (PTPlD) from ~-[~~S]Met-labeled 293 cell transfectants was analyzed with or without stimulation by EGF. On EGF stimulation, the phosphorylation level of SH-PTP2 was elevated and protein tyrosine phosphatase activity was increased slightly (1.2-fold). The contribution of phosphorylation and SH2 domain binding to this increase in phosphatase activity is unclear. Experiments are currently under way in our labs to clarify these issues.