Activation of the SH2-containing Protein Tyrosine Phosphatase, SH-PTPB, by Phosphotyrosine-containing Peptides Derived from Insulin Receptor Substrate-l*

The cytoplasmic insulin receptor substrate-1 (IRS-l), which is multiply phosphorylated in vivo on tyrosine residues, is a known binding protein for the tandem src homology 2 (SH2) domain-containing protein tyrosine phosphatase, SH-PTP2. Eleven phosphotyrosyl (pW peptides from IRS-1 were screened for allosteric activation of SH-PTP2 phosphatase activity toward phosphorylated, reduced, carboxyamidomethylated, and maley- lated-lysozyme. Peptides IRS-lpY896, IRS-lpY1172, and IRS-lpY1222 showed up to SO-fold acceleration of de- phosphorylation. Analyses of Arg to Lys mutants in either or both SH2 domains indicate that both the N-ter-mid (N-SH2) and C-terminal (C-SH2) domains function in allosteric activation. Direct determination by surface plasmon resonance of the dissociation constants between pY peptides and glutathione S-transferase fusions to N-SH2 and C-SH2 domains reveals a 240-fold preference of the N-SH2 domain (compared with the C-SH2 domain) for IRS-lpY1172. The N-SH2 domain prefers IRS-lpY1172>IRS-lpYB95* IRS-lpY1222, whereas C-SH2 domain prefers IRS-lpY1222>IRS-lpY89S>IRS-lpYll72. These data suggest that each SH2 domain can bind to a distinct pY sequence of multiply phosphorylated protein substrates such as IRS-1, while activating hydrolysis at a third pY sequence bound in the SH-PTP2 active site. In addition, proteolysis and truncation studies reveal an autoregulatory function for the C-terminal region of SH- PTP2. Limited tryptic cleavage within the C terminus results in 27-fold activation of protein tyrosine phosphatase activity. The activated tryptic fragment cannot be M h e r activated by pY peptide binding to the SH2 domains indicating that autoregulatory functions of the SH2 domains are dependent on the C-terminal region. These data suggest that multiple levels for control of SH-PTP2 enzymatic activity may exist in vitro and in vivo. preparations of SH-PTP2 prote- significant in relative SH-PTP2.

Many polypeptide growth factors signal via receptors with intrinsic protein tyrosine kinase activity (1). In recent years, much progress has been made in defining early events in receptor protein tyrosine kinase signaling (2,3). Upon ligand addition, receptor protein tyrosine kinases dimerize, are enzymatically activated, and "trans-phosphorylate" at multiple sites (1) in their cytoplasmic domains. These tyrosyl phosphorylation sites serve as docking points to recruit secondary signaling molecules, many of which contain src homology-2 (SH2)' domains and are also receptor protein tyrosine kinase substrates (1). SH2 domains, first identified in src-family protein tyrosine kinases, are regions of approximately 100 amino acids which bind with high affinity (e.g. to M) to specific phosphotyrosyl (pY) peptides (3)(4)(5)(6)(7)(8)(9). The specificities of SH2 domains for particular phosphorylation sequences have been demonstrated in vitro by direct assays using SH2 domain and pY peptides (6)(7)(8) and in vivo by the finding that different pY residues on receptor protein tyrosine kinases associate with distinct downstream SH2 domain-containing proteins (3)(4)(5).
Our goal is to clarify the inter-and intramolecular mechanisms of regulation of SH-PTPase activity. We reported the purification of full-length recombinant human SH-PTP2 produced in Escherichia coli and the characterization of its enzymological properties (24). We noticed that SH-PTP2 exhibits non-Michaelis-Menten kinetics toward a pY peptide encompassing l j~~~~~ of the human PDGFR (PDGFRpY1009), which we (25) and others (26) found to be the high affinity binding site for SH-PTP2 on the PDGFR. Furthermore, addition of the synthetic pY peptide PDGFRpY1009, but not other pY peptides based on the PDGFR, resulted in activation of SH-PTP2 enzymatic activity (25). We also reported that a glutathione S-transferase (GST) fusion protein of the N-terminal SH2 domain of SH-PTP2 (N-SH2) binds directly to the PDGFR in vitro (25). Out of seven autophosphorylation sites of PDGFR, this fusion protein binds with high affinity to only one pY peptide (PDG-FRpY1009) (27). These data suggest that SH-PTP2 binds to PDGFR via its N-SH2 domain.
In the present study we assess the respective contributions of the N-and C-terminal SH2 domains of SH-PTP2 in the recognition of specific pY peptides from IRS-1 and their ability to activate the SH-PTP2 catalytic domain. IRS-1 is directly phosphorylated at many tyrosine sites by the insulin receptor in vitro (21)(22)(23) and in vivo (22,23). We demonstrate here that three pY peptides derived from IRS-1 are capable of specific binding to SH-PTP2 SH2 domains and of evoking up to 50-fold catalytic activation of SH-PTP2. Occupancy of either SH2 domain is communicated to the PTPase active site. In addition to regulation via its SH2 domains, we define a region of the SH-PTP2 C terminus that also has a regulatory function.

EXPERIMENTAL PROCEDURES
Materials-E. coli cells harboring a n expression plasmid for a GST fusion protein of amino acids 101-268 of SH-PTP2 (GST-C-SH2) was reported previously (27).
To generate pET-R32WR138K, a three way ligation was performed to join the 0.5-kb SphI-BgZI fragment from pET-R32K (corresponding to 0.3-kb upstream of the structural gene and encoding amino acid residues 1-73 of SH-PTP2 with a substitution of Lys for Ar&'), the 1.8-kb BglI-Sal1 fragment from PET-R138K (encoding amino acid residues 74-593 of SH-PTP2 with a substitution of Lys for Arg13') and the down- Expression and Purification of SH2 Domain GST Fusion Proteins-E. coli strain DH5a was transformed with pGEX-2N and pGEX-2NR32K, respectively, and induced to produce GST fusion proteins to the N-SH2 domain (GST-N-SH2) and the N-SH2 domain with substitution of A r $ ' to Lys (GST-N-SH2R32KX respectively. Expression and purification of these GST fusion proteins and GST-C-SH2 were performed by the method of Payne et al. (6). GST fusion proteins were greater than 90% pure, as assessed by SDS-PAGE.
E. coli cells expressing PC61 were cultured, induced, and lysed. After lysis, the supernatant was loaded onto a Q-Sepharose Fast Flow (Sigma) column (11 x 2.5 cm) equilibrated with 100 ml of 10 m Tris-HCl, pH 7.5, containing 1 nw EDTA and 10 nw 2-mercaptoethanol (bufTer A). The column was washed with 150 ml of buffer A and a 0-500 II~M NaCl concentration gradient was developed in 400 ml of buffer A at 1 mUmin. Fractions eluting from 200 to 250 nw NaCl were pooled, and ammonium sulfate was added to 40% saturation. The supernatant was loaded onto a phenyl-Sepharose column (13 x 2.5 cm) equilibrated with buffer A containing 40% saturated ammonium sulfate. The column was washed with the same buffer and activity was eluted using a n ammonium sulfate concentration gradient, 40% to 0% in 400 ml of buffer A at 0.5 mumin. The active fractions eluted at approximately 4% saturated ammonium sulfate and were desalted on a prepacked Sephadex G-25 column (PD-10, Pharmacia) equilibrated with buffer A, and loaded onto a Mono Q HR 10/10 (Pharmacia) column equilibrated with buffer A. The latter column was developed with a gradient of 0-250 m NaCl in 250 ml of buffer A a t 1 mumin. Fractions eluting from 200 to 250 m NaCl represent over 90% pure AC61, as assessed by SDS-PAGE. These fractions were pooled and concentrated using a Centriprep-10 (Amicon) and stored at -80 "C in the presence of 33% (v/v) glycerol.
Limited Digestion of Wild l)pe SH-PTP2 with Dypsin or Chymotrypsin-Purified, wild type SH-PTP2 (440 pg/ml) was digested with 0.88 pg/ml trypsin a t 22 "C for 15 min or 4.4 pg/ml chymotrypsin a t 37 "C for 4 h in 100 m Tris-HC1, pH 7.6, containing 0.1 m CaCl,. These reactions were quenched by the addition of soybean trypsin inhibitor to a final concentration of 40 pg/ml. In control experiments, proteases were preincubated in the same buffer with trypsin inhibitor at 37 "C for 15 min and added to wild type SH-PTP2.
Assay for PTPase Actiuity-RCM-lysozyme (Life Technologies, Inc.) was phosphorylated on tyrosine (28) using recombinant v-Ab1 (Oncogene Science, Inc) according to the manufacturer's instructions, except that ATP was present at 4 m and [g2PlATP was added to 500 pCi/ml. The typical specific radioactivity obtained was 1500 countdmidpmol.
Activation of SH-PTP2 by m~ HEPES, pH 7.4, 150 m~ NaCl, 100 p g h l bovine serum albumin, 5 m~ EDTA, and 10 m~ dithiothreitol). Phosphate release was measured using a charcoal binding assay (24), and reaction velocity was quantitated from the counts in the supernatant and the specific radioactivity of RCM-lysozyme. Under these conditions, phosphate release is linear with respect to the time of incubation (24).
Phosphopeptide Synthesis-Phosphopeptides were synthesized as described in Piccione et al. (28) using the methodology of Kitas et al. (29). Peptide residues at the +1, +2, and +3 positions relative to phosphotyrosine are most important for conferring SH2 domain specificity (4,8,9,30). Crystallographic structural analyses of lck and src SH2 domaidphosphopeptide complexes suggest that important contacts may occur with peptide positions ranging from -2 to +4 position relative to phosphotyrosine as well (31,32). Most of the peptides contain 11 amino acids, with 3 residues N-terminal and 7 residues C-terminal to phosphotyrosine. In some case an extra residue was added or deleted for synthetic purposes. The sequence of the phosphopeptides used were as follows, with pY indicating the phosphorylated tyrosine: IRS-lpY147, EDLSpYDTGPGPA, IRS-lpY460, LSNpYICMGGKG; IRS-lpY546, IEEpYTEMMPAA, IRS-lpY608, DDGpYMPMSPGV; IRS-lpY628, GNGDpYMPMSPKS; IRS-lpY658, PNGpYMMMSPSG; IRS-lpY727, TGDpYMNMSPVG IRS-lpY895, SPGEpYVNIEFGS; IRS-lpY987, RGDpYMTMQIG; IRS-lpY1172, SLNpYIDLDLVK, IRS-lpY1222, LST-pYASINFQK. m n i t y Measurement Using Plasmon Resonance Analysis-Kd values for pY peptides and SH2 domains were measured by plasmon resonance using a BIAcore (Pharmacia) (33) instrument. The conditions for immobilization of phosphopeptides have been described (6,7). For IRS-lpY895, a Lys residue was attached to a-amino group of the N-terminal Thr to create a reactive €-amino group for immobilization. Kd values were calculated from the association (k,) and dissociation (k,) rate constants or obtained from equilibrium experiments (6,7). For the N-SH2 domain, a rank ordering of Kd values were also measured by a BIAcore peptide competition assay (6,7), using an IRS-lpYll72-immobilized surface.
Other Methods-Protein concentrations were determined by Bradford assay, using a commercially available kit (Bio-Rad) and bovine serum albumin as standard (34). SDS-PAGE was carried out as described by Laemmli (35). N-terminal sequencing was performed by the Edman method (36) using an automated gas-phase sequenator (Applied Biosystems).

RESULTS
Rat and human IRS-1 have been cloned (37, 38) and show high identity in amino acid sequence, especially regarding the position of tyrosines and their flanking sequences. Rat IRS-1 has 34 Tyr residues (37) and human IRS-1 has 32 Tyr residues (38), and all 32 Tyr residues of human IRS-1 are conserved in rat IRS-1. At 27 of the 32 conserved Tyr sites, the sequences of the 3 amino acid residues immediately following each Tyr are identical; such pYxxX motifs frequently represent docking sequences for SH2 domains (31,32,39). In this study, we used ll synthetic peptides corresponding to rat IRS-1 sequences, including the known phosphorylation sites (231, to determine the effects of SH2 occupancy on SH-PTP2 PTPase activity. At 80 p~ concentration, three out of the 11 IRS-1 derived pY peptides (IRS-lpY895, IRS-lpY1172, and IRS-lpY1222) demonstrated activation of SH-PTP2 PTPase activity toward pY-RCM-lysozyme (Fig. 1). We chose these three pY peptides for studies of the dose dependence of activation (Fig. 2). Two other pY peptides (IRS-lpY546 and IRS-lpY727) that did not activate served as negative controls. IRS-lpY1172 produced a biphasic activation profile, in which SH-PTP2 was activated 15-fold at lower concentration of pY peptide (ED,, of 5-10 ) 1~) and 25-fold further activation (total of 40-fold activation) was observed at higher concentrations of pY peptide (ED,, of 200 PM). Under these conditions, IRS-lpY895 and IRS-lpY1222 demonstrated monophasic activation (ED,, of 200 p~). Interestingly, IRS-lpY1222 evoked higher maximal fold activation (60-fold) than IRS-lpY1172 or IRS-lpY895 (40-fold). IRS-lpY546 and IRS-lpY727 showed no significant effect within the concentration range tested. It was unclear why IRS-lpY1172 showed biphasic activation; we suspected that occupancy of either the N-and C-SH2 domains by pY peptides might provide allosteric activation and that IRS-lpY1172 might have a different affinity for each SH2 domain. Thus, at lower concentrations, IRS-lpY1172 could bind first to one SH2 domain and show activation to some extent, whereas at higher concentrations, it would bind to the second SH2 domain, yielding further activation. By contrast IRS-lpY895 and IRS-lpY1222 might bind to only one SH2 domain or have comparable lower affinity for both SH2 domains. To test this hypothesis, we analyzed SH2 domain recognition of IRS-1 pY peptides in two ways: (i) by examining the effects of SH2 domain mutations on SH-PTP2 activity and (ii) by direct determination of the Kd between pY peptides and each SH2 domain of SH-PTP2.

SH2 Domain Mutants: Effects on Allosteric
Activation-Structural studies have established that a specific Arg residue in the SH2 domains of src, abl, and lck stabilizes the negatively charged phosphotyrosine group of bound pY ligands (31,32,39,40). An Arg to Lys mutation at this conserved residue leads to significantly weaker binding of pY peptides (41). The Arg to Lys mutations were made in SH-PTP2 in the N-SH2 domain at k g 2 , in the C-SH2 domain at Arg13*, and in both SH2 domains.
The three mutant enzymes were expressed in E. coli, purified to homogeneity, and their kinetic properties were compared with the wild type enzyme.
The double Arg to Lys substitution mutant, R32WR138K, showed no catalytic activation up to 100 p~ IRS-lpY1172, whereas wild type SH-FTP2 showed 25-30-fold activation at 100 PM IRS-lpY1172 (Fig. 31, confirming that IRS-lpY1172 activates SH-PTP2 via its SH2 domain(s). Activation of R32W R138K by IRS-lpY1172 at concentrations greater than 200 p may result from residual affinity of the SH2 domains for pY peptides.
The single R138K substitution mutant in the C-SH2 domain showed PTPase domain activation with a n ED,, of 5-10 p~, indicating that the wild type N-SH2 domain in R138K was responsible for high affinity activation. In the R32K N-SH2 domain mutant, no activation by IRS-lpY1172 was detected at concentrations less than 100 p~. Along with the results on the double substitution mutant, these data indicate that the N-SH2 domain is responsible for high affinity binding and activation.
However, at higher concentrations of IRS-lpY1172, the R32K mutant could be allosterically activated. Interestingly, at these higher concentrations (>lo0 p) R32K displayed a steep activation curve, with maximal activation at -2 m~ IRS-lpY1172, whereas R138K or R32WR138K showed a more shallow activation profile. These data suggest that the C-SH2 domain can be occupied at higher concentration of IRS-lpY1172 and activate the catalytic domain.
Direct Analysis of Kd of SH-PTP2 SH2 Domains for pY Peptides by Surface Plasmon Resonance-Purified GST fusion proteins of the Nor C-SH2 domains of SH-PTP2 were used to measure the Kd for binding to immobilized IRS-lpY895, IRS-lpY1172, or IRS-lpY1222, by surface plasmon resonance analysis (6,7,33) (Table I). Typical sensorgrams are shown in Fig.  4 for binding of GST-N-SH2 to IRS-lpY1172 (Fig. 4A) or GST-C-SH2 to IRS-lpY1222 (Fig. 4B). Kd values were calculated from rate constants of association (k,) and of dissociation (k,) or measured from response units at equilibrium (ReJ and compared by the peptide competition assay (6) as indicated in Table  I. The measured affinities of the three pY peptides for both SH2 domains varied over a wide range.
Most notably, IRS-lpY1172 shows high affinity (14 m) for the N-SH2 domain, some 240-fold higher than its affinity for the C-SH2 domain. These results are qualitatively consistent with the mutagenesis studies described above. The fusion protein containing the R32K mutation, GST-N-SH2R32K, showed about 50-fold lower affinity to the IRS-lpY1172 surface (640 versus 14 m) than the wild type N-SH2 domain.
Although the IRS-lpY895 peptide shows almost 30-fold lower affinity for the N-SH2 domain (390 m) than IRS-lpY1172, its Kd for the C-SH2 domain (310 nM) is some 10-fold less than IRS-lpY1172 for this SH2 domain. In relative terms, the IRS-lpY895 preference between N-and C-SH2 domains is  1:l versus 240:l for IRS-lpY1172, thus explaining the monophasic activation curve seen for IRS-lpY895 activation of PTPase activity (Fig. 2). The third IRS-1 pY peptide, IRS-lpY1222, shows 60-fold lower affinity for the N-SH2 domain (900 m) than IRS-lpY1172, whereas its Kd for the C-SH2 domain (110 nM) is some 30-fold less than IRS-lpY1172 for this SH2 domain. Compared with IRS-1 pY1172, IRS-lpY1222 displayed the converse specificity, with a 1:8 preference for N-versus C-SH2 domains. Given the relative magnitudes of the Kd values of IRS-lpY1222 for the N-SH2 and C-SH2 domains and the ED,, for activation of PTPase activity, it is likely that activation by IRS-lpY1222 arises via the C-SH2 domain, although it is possible that the 8-fold difference in Kd is small enough that biphasic activation by IRS-lpY1222 appears to be monophasic.

Role of the C-terminal Region of SH-pTP2 in Regulation of
Catalytic Activity-In addition to the two tandem SH2 domains upstream of the catalytic PTPase domain, SH-PTP2 and SH-PTPl have additional amino acids (69 and 81 amino acids, respectively), downstream of the catalytic core (10, 11). Activation of SH-PTP1 by C-terminal truncation, releasing 41 amino acids, was reported previously (42). We were therefore interested in determining the effect of the SH-ITP2 C terminus on PTP activity and its possible relationship to regulation via the SH2 domains. Full-length, wild type SH-PTP2 was subjected to limited digestion with either trypsin or chymotrypsin. Chymotrypsin digestion yielded a stable fragment about 3.5 kDa smaller than the wild type enzyme whereas tryptic digestion yielded a stable fragment about 4.4 kDa shorter (Fig. 5A). N-terminal sequence analyses of trypsin-and chymotrypsin-treated SH-PTP2 revealed that both retain the original N-terminal sequence (data not shown). Thus, both proteases cleave within the C-terminal region. Inspection of the SH-PTP2 protein sequence reveals the potential cleavage sites noted in Fig. 5B. The tryptic fragment is approximately 1 kDa larger than the chymotryptic fragment (Fig. 5A), consistent with the suggested sites of cleavage (Fig.  5B); however, a dramatic difference in basal PTPase specific activity was observed. The chymotryptic fragment possessed the specific activity of the wild type enzyme, but the specific activity of the trypsin-treated fragment was elevated 27-fold, indicating a disruption of autoregulation by tryptic cleavage (Table 11). When these two fragments were assayed for activation by IRS-lpY1172, the chymotryptic fragment was activated 18-fold, comparable to the 30-fold activation with full-length enzyme. However, there was no further effect of IRS-lpY1172 addition to the trypsin-treated fragment, (already activated 27-fold) suggesting that the trypsin-cleaved enzyme was already fully activated.
We also analyzed the enzymological properties of an SH-PTP2 C-terminal deletion mutant (AC61). As shown in Fig. 5A, the purified AC61 protein had the expected molecular mass, apparently 5 kDa less than full-length enzyme. Surprisingly, type basal specific activity and the capacity for marked activathe AC61 protein, which is only slightly smaller than the (action (19-fold) by IRS-lpY1172. tivated) protein generated by tryptic cleavage, retained wild Three points emerge. First, proteolysis by trypsin at a spe-

Activation of SH-PTP2 by C-terminal truncation and pY peptide
Control preparations (digestion of SH-PTP2 in the presence of protease inhibitors) show no significant differences in relative activity from wild type SH-PTP2. cific region, probably a t Lys536, AT$^^, or Lys538, removes negative autoregulation. Second, both larger (chymotrypsin proteolysis) and smaller (AC61) truncated enzymes do retain full negative autoregulation. Third, once the enzyme is activated by trypsin cleavage at the C terminus, it cannot be further activated by SH2 domain occupancy by pY ligands.

DISCUSSION
PTPases are of interest both for their catalytic mechanism, involving a covalent S-phosphocysteinyl enzyme intermediate (4346) and for their biological role in the regulation of signal transduction. The SH2 domain-containing FTPases, SH-PTP2, examined here, and SH-PTP1, are of particular enzymological interest because they contain three domains capable of recognizing pY proteins: two high affinity SH2 domains and the lower affinity catalytic domain, which possesses an irreversible hydrolytic function. In the evaluation of the possible biological roles of SH-PTPS, some attention to date has focused on assessing the binding selectivity and specificity of the two SH2 domains for pY peptides.
Songyang et al. (9) screened a pY peptide library with GST fusions of the N-SH2 or C-SH2 domains of SH-PTP2 and observed binding only to the GST-N-SH2 with a sequence preference of pY-W-X-VII. More recently, based on the results of peptide competition assays, Case et al. (27) suggested the refinement pY-V/TA"X-VILII for N-SH2 domain binding; that is, a preference for 0-branched amino acids at the +1 and hydrophobic residues at the +3 positions. Previously, we did not isolate ligands to the C-SH2 domain, possibly due to the use of a shorter construct of GST-C-SH2 (9,21). We have previously reported a functional assay to assess the occupancy of SH2 domains by pY ligand-coupled activation of full-length SH-FTP2. Using pY peptides from the PDGFR, we noted that PDG-FRpY1009 (SVLpYTAVQPNE) stimulated catalytic activity, measured as release of 32Pi from 32P-pY-RCM-lysozyme (25). This biochemical observation is likely to have important implications for SH-PTP2 function in vivo, since the PDGFR is a likely relevant in vivo partner of this PTPase (16, 17,21,22), and we (25) and others (26) have identified pY1009 as the high affinity site for SH-PTP2 binding to the activated PDGFR.
To study the relationship between SH2 domain recognition and allosteric activation, we analyzed pY peptides comprising seven of the above eight in vitro phosphorylation sites and four others derived from IRS-1, IRS-lpY147, IRS-lpY546, IRS-lpY658, and IRS-lpY727 for their ability to stimulate the PTPase activity of full-length SH-P"2. Indeed, only three pY peptides IRS-lpY895, IRS-lpY1172, and IRS-lpY1222 ofthe 11 studied showed stimulation of PTPase activity. This simulation was dramatic, up to 40-60-fold increase of V,,, even more pronounced than the 10-15-fold effect seen with PDG FRpY1009 (25).
To further analyze the contributions of the individual SH2 domains of SH-F'TP2 to allosteric activation, mutagenesis of the conserved arginines, A r $ z (N-SH2) and (C-SH2) in either or both of the SH2 domains was undertaken. The data on the R32K/R138K double mutant are particularly compelling in that the allosteric activation of PTPase activity by IRS-lpY1172 is reduced by two orders of magnitude, proving that one or both SH2 domains mediate(s) SH-PTP2 activation. Moreover, these data, combined with our previous demonstration (25) that activation was dependent on the phosphorylation of the tyrosine-containing peptides, strongly suggests that activation is mediated via phosphotyrosine-dependent binding to the SH2 domains.
The R32K single mutant is -50-fold less responsive than wild type, whereas the R138K mutant retains some high affinity activation capacity (6-8-fold). These results suggest that the N-SH2 domain is indeed the high affinity IRS-lpYll72-binding site and that occupancy of the N-SH2 domain alone (at low concentrations of IRS-lpY1172) is sufficient to evoke some level of SH-PTP2 activation. However, several lines of evidence also indicate that occupancy of the C-SH2 domain can also lead to allosteric activation.
IRS-lpY1172 shows a biphasic dose-dependence activation curve, with maximal activation occurring at millimolar concentrations of IRS-lpY1172. This observation is most easily explained by binding of this peptide to the N-SH2 at low concentrations yielding partial SH-PTP2 activation followed by C-SH2 domain occupancy at higher concentrations leading to full activation of the enzyme. Consistent with this notion, the R32K mutant, which lacks a functional N-SH2 domain, can be activated by IRS-lpY1172 but only at high concentrations of the peptide (maximal activation approached at 2 m~ IRS-lpY1172). Likewise, the finding that R138K and R32m138K have weakened ability to elicit full activation at higher IRS-lpY1172 concentration implies that the C-SH2 domain functions in activation at higher concentration of IRS-lpY1172. The mutagenesis data indicate that both of the SH2 domains participate in allosteric activation of the PTPase domain. Moreover, from a biological standpoint, these data raise the possibility that upon interaction with different target proteins containing distinct pY peptide sequences different extents of PTP activation may occur in vivo. Such differences could have important consequences for downstream signaling.
Direct determination by surface plasmon resonance of the affinity of the N-SH2 and C-SH2 domains for the three activating IRS-1 pY peptides revealed additional correlations. IRS-lpY1172 (pYIDL) with a Kd of 14 n~ is a high affinity ligand for the N-SH2 domain; some 50-fold loss in affinity is observed with the GST-N-SH2R32K mutant. The IRS-lpY895 sequence (pYVNI) is recognized 30-fold less well and the pYASI sequence of IRS-lpY1222 some 60-fold less well by the N-SH2 domain. Our results differ somewhat from those of Case et al. (27), who reported that IRS-lpY1172 binds 4-5-fold more tightly to the N-SH2 domain than IRS-lpY895 (compared with the 30-fold preference observed in this study; see Table I). This discrepancy is probably due to methodological differences between the two studies. In particular, it should be noted that the GST-N-SH2 domain fusion protein used by Case et al. (27) was refolded during purification. This process might lead to slight decreases in the binding ability of the fusion protein.
Analyses of the C-SH2 domain reveal data supporting its ability as an isolated domain to selectively bind pY sequences. binding event would be expected to bring the pY1222 or pY895 sites into activated SH-PTP2 but not to maximal levels. However, this primary close proximity to the C-SH2 domain leading to binding of the C-SH2 domain to one of these sites with full activation of SH-PTP2 enzymatic activity. The active site of bound SH-PTP2 may then be juxtaposed to additional tyrosyl phosphorylation sites on IRS-1, possibly including the binding sites of signal transducing proteins such as GRB2, PIBK, or others. SH-PTP2 may regulate signal transduction by dephosphorylating these sites or phosphotyrosyl residues of associated protein substrates.
More importantly, the C-SH2 domain displays substantially different binding preferences relative to the N-SH2 domain.
Both IRS-lpY895 and IRS-lpY1222 bind to the C-SH2 domain 10-30-fold more tightly than IRS-lpY1172 with Kd values of 310-110 m, respectively. This reversal of selectivity indicates that the N-SH2 and C-SH2 domains of SH-PTP2 recognize distinct pY sequences.
Together with the SH2 domain mutant results, these data suggest that the N-SH2 domain is responsible for sensitive allosteric activation by IRS-lpY1172, whereas the C-SH2 domain is responsible for the 60-fold allosteric activation by IRS-lpY1222 and contributes to activation by IRS-lpY895. This gives rise to the prospect that recognition and dephosphorylation of a multiphosphorylated IRS-1 by SH-PTP2 could involve allosteric engagement of two distinct pY sequences (e.g. IRS-lpY1172 in the N-SH2 domain and IRS-lpY1222 in the C-SH2 domain, given the range of IRS-1 peptides examined in this study), while a third pY sequence (either on IRS-1 or a distinct substrate) undergoes allosterically activated hydrolysis in the catalytic site (see Fig. 6).
It should also be noted (Fig. 2) that maximal activation (at saturation) by IRS-lpY1222 (60-fold) is greater than that produced by IRS-lpY895 and IRS-lpY1172 (40-fold). This suggests that different peptides, when bound to the same SH2 domain, may result in subtle differences in SH2 structure which can then have important regulatory consequences.
Prior work on SH-PTP1 (PTPlC) indicated that proteolytic cleavage of 41 amino acids from its C-terminal region enhanced basal enzymatic activity 15-fold (42). We observed similar behavior with SH-PTP2 upon limited exposure to trypsin, in which a -4.4-kDa fragment is cleaved from the C terminus ( Fig. 5 and Table 11) to yield a catalytic fragment with a 27-fold higher V, , than the full-length enzyme. Notably, this truncated, activated form of SH-PTP2 shows no further activation by pY ligand binding to either SH2 domain. However, both a slightly larger fragment of SH-PTP2 generated by limited chymotryptic digestion and a slightly smaller fragment (PC611 generated by recombinant DNA techniques have basal level PTPase activity that is fully activatable by pY ligand occupancy of either SH2 domain. Similar results have been observed with SH-PTP1 using two C-terminal deletion mutants AC35 and AC60 that were constructed and purified to homogeneity. Although activating pY peptides have not yet been found for SH-PTP1, SH-PTPlAC35 shows 41-fold higher PTPase activity toward pY-RCM-ly-sozyme than full-length SH-PTP1 while SH-PTPlAC60 shows only 2-fold higher activity than full-length SH-PTP1. ' We have also observed that SH-PTP2 mutants in which both SH2 domains have been deleted (Se?28-Ar$93) or in which both SH2 domains and the C-terminal61 amino acids have been deleted (SeP28-G1~53Z) possess at least 200-fold higher para-nitrophenyl phosphatase a~t i v i t y .~ These data suggest a model in which, in the absence of pY peptides, either or both SH2 domains inhibit PTPase activity and that the C-terminal region participates in autoinhibition by SH2 domain(s).
Therefore, it is likely that a region of SH-PTP2 from Gln533 to the C-terminal end of the chymotryptic fragment (estimated to be between residues 546-552) has the capacity to modulate PTPase activity. To clarify the C-terminal modulation mechanism of PTPase activity, further studies of the SH-PTP2 C terminus are required. Inspection of the amino acid sequence reveals several potential regulatory elements including a cluster of charged residues from Lys534 to Lys538 (KSKRK) as well as proline-rich region (6 prolines in the 10 residue stretch from Prossg to Pro5' ?. A potential N-SH2 domain preferred sequence is also present (YTNI, ' &P' to Ile545). Indeed, we have recently found that 'I'yrX2 is the major site of in vivo and in vitro tyrosyl phosphorylation of SH-PTP2 in response to PDGF stimulation.
Moreover, when phosphorylated, this site serves as a binding site for the adapter GRBB bound to the Ras exchange protein mSOS in vitro and in vivo. 4 Thus, in vivo phosphorylation and/or GRBB binding might also influence activity of the enzyme, in addition to constitutive phosphorylation at Ser and Thr (16, 17,21,52). Since SH-PTP2 becomes tyrosyl phosphorylated in response to some (PDGF, epidermal growth factor, hepatocyte growth factor), but not all (insulin) growth factors (16,17,21,22,53), our studies suggest additional potentially relevant levels of control of SH-PTP2 activity.
It is unclear why there is a substantial difference between the Kd for SH2 domain binding measured with the BIAcore and the ED,, for PTPase allosteric activation. In each case, the ED,, is -1000-fold higher than the measured Kd value. The Kd of IRS-lpY1172 for the N-SH2 domain is 14 n M which is about 1000-fold lower than the ED,, of 5-10 p~ for the allosteric effects of pY peptides on PTPase Vna. Nevertheless, the loss of both peptide binding and peptide-mediated allosteric activation in the R32K mutant of SH-PTP2 clearly indicates that activation is mediated by pY peptide binding to the N-SH2 domain. Similar discrepancies between binding and activation have been observed in other studies. It has been reported that IRS-1 pY peptides ( p Y " ) bind to the SH2 domains of the p85 subunit of PI3K (47,48) and activate the lipid kinase (54,55). The ED,, for kinase activation is also 500-15,000-fold higher than the Kd values for phosphopeptide binding to p85 (7,54,56). It is possible that the observed Kd value is artificially low due to rebinding of the SH2 domain; however, it is more likely that another part of SH-PTP2 (or p85 of PI3K) present in the full-length molecule greatly attenuates the SH2 domain recognition. The SH2 domains may physically associate with each other, and this interaction might alter SH2 domain affinity for various pY peptides. The C-terminal region may also participate in such intramolecular physical interactions to modulate conformational changes and enzyme activity. We would predict that the affinity for phosphopeptide binding by full-length SH-PTP2 is substantially lower than that measured by direct binding to GST fusions of the isolated N-SH2 or C-SH2 domains.
The precise molecular mechanism of allosteric activation upon occupancy of either or both of the SH2 domains as well as how these effects are transmitted to the active site remain to be resolved. Future studies will be directed toward elucidating the molecular details of both inter-and intramolecular regulation of SH-PTP2 activity. In particular, it will be important to clarify whether and how these regulatory mechanisms influence specific steps in PTP catalysis such as formation or breakdown of the covalent S-phosphocysteinyl enzyme intermediate.
In summary, our studies establish that both SH2 domains of SH-PTP2 function in allosteric activation, that the C-terminal region is autoregulatory, and that activation via SH2 domains is related to autoregulation of activity by the C-terminal region. The selective activation by specific IRS-1 pY sequences along with known behavior of IRS-1 as a ligand of SH-PTP2 suggest this will be an interesting protein pair with which to assess SH-PTP2 function.