The Role of C-terminal Tyrosine Phosphorylation in the Regulation of SHP-1 Explored via Expressed Protein Ligation*

The protein-tyrosine phosphatase SHP-1 plays a variety of roles in the “negative” regulation of cell signaling. The molecular basis for the regulation of SHP-1 is incompletely understood. Whereas SHP-1 has previously been shown to be phosphorylated on two tail tyrosine residues (Tyr536and Tyr564) by several protein-tyrosine kinases, the effects of these phosphorylation events have been difficult to address because of the intrinsic instability of the linkages within a protein-tyrosine phosphatase. Using expressed protein ligation, we have generated semisynthetic SHP-1 proteins containing phosphotyrosine mimetics at the Tyr536 and Tyr564 sites. Two phosphonate analogues were installed, phosphonomethylenephenylalanine (Pmp) and difluorophosphonomethylenephenylalanine (F2Pmp). Incorporation of Pmp at the 536 site led to 4-fold stimulation of the SHP-1 tyrosine phosphatase activity whereas incorporation at the 564 site led to no effect. Incorporation of F2Pmp at the 536 site led to 8-fold stimulation of the SHP-1 tyrosine phosphatase activity and 1.6-fold at the 564 site. A combination of size exclusion chromatography, phosphotyrosine peptide stimulation studies, and site-directed mutagenesis led to the structural model in which tyrosine phosphorylation at the 536 site engages the N-Src homology 2 domain in an intramolecular fashion relieving basal inhibition. In contrast, tyrosine phosphorylation at the 564 site has the potential to engage the C-Src homology 2 domain intramolecularly, which can modestly and indirectly influence catalytic activity. The finding that phosphonate modification at each of the 536 and 564 sites can promote interaction with the Grb2 adaptor protein indicates that the intramolecular interactions fostered by post-translational modifications of tyrosine are not energetically strong and susceptible to intermolecular competition.

The protein-tyrosine phosphatase SHP-1 plays a variety of roles in the "negative" regulation of cell signaling. The molecular basis for the regulation of SHP-1 is incompletely understood. Whereas SHP-1 has previously been shown to be phosphorylated on two tail tyrosine residues (Tyr 536 and Tyr 564 ) by several protein-tyrosine kinases, the effects of these phosphorylation events have been difficult to address because of the intrinsic instability of the linkages within a protein-tyrosine phosphatase. Using expressed protein ligation, we have generated semisynthetic SHP-1 proteins containing phosphotyrosine mimetics at the Tyr 536 and Tyr 564 sites. Two phosphonate analogues were installed, phosphonomethylenephenylalanine (Pmp) and difluorophosphonomethylenephenylalanine (F 2 Pmp). Incorporation of Pmp at the 536 site led to 4-fold stimulation of the SHP-1 tyrosine phosphatase activity whereas incorporation at the 564 site led to no effect. Incorporation of F 2 Pmp at the 536 site led to 8-fold stimulation of the SHP-1 tyrosine phosphatase activity and 1.6-fold at the 564 site. A combination of size exclusion chromatography, phosphotyrosine peptide stimulation studies, and site-directed mutagenesis led to the structural model in which tyrosine phosphorylation at the 536 site engages the N-Src homology 2 domain in an intramolecular fashion relieving basal inhibition. In contrast, tyrosine phosphorylation at the 564 site has the potential to engage the C-Src homology 2 domain intramolecularly, which can modestly and indirectly influence catalytic activity. The finding that phosphonate modification at each of the 536 and 564 sites can promote interaction with the Grb2 adaptor protein indicates that the intramolecular interactions fostered by post-translational modifications of tyrosine are not energetically strong and susceptible to intermolecular competition.
SHP-1 is a 68-kDa protein composed of two SH2 domains, a tyrosine phosphatase catalytic domain and a flexible C-terminal domain which has been proposed to play a regulatory role (Fig. 1A). There is a high resolution crystal structure of the SHP-1 catalytic domain indicating the classical PTPase fold (10). These enzymes have a highly conserved cysteine residue that serves as the catalytic nucleophile, generating a phosphocysteine covalent intermediate (11). The phosphoenzyme intermediate is then hydrolyzed producing inorganic phosphate. Whereas the authentic physiologic protein substrates of SHP1 have not been determined with certainty, several proposed phosphoprotein targets include Lyn, Syk, BLNK/SLP-65, Lck, ZAP-70, phosphatidylinositol 3-kinase, SLP-76, interleukin 2 receptor, IRF1, and the interferon consensus sequence-binding protein (1,12,13). Typically phosphopeptides or para-nitrophenol phosphate (pNPP) have been used to study the catalytic mechanism and regulation of SHP-1 in vitro (14).
The mechanisms of the regulation of SHP-1 are incompletely understood. No high resolution structure has been reported for the full-length protein. The two SH2 domains may be important in recruiting phosphotyrosine-containing proteins, controlling substrate specificity, and cellular localization of the enzyme. In addition, the N-terminal SH2 domain appears to play a role as a negative regulator of SHP-1 catalytic activity by directly binding to the SHP-1 catalytic domain (15). This inhibition can be relieved by phosphotyrosine containing peptides with appropriate sequences and possibly phospholipids (15,16). Two C-terminal tyrosine phosphorylation sites at Tyr 536 and Tyr 564 have been mapped (Fig. 1A) and possible tyrosine kinases that catalyze these reactions include Lck, Abl, and the insulin receptor tyrosine kinase (7,(17)(18)(19). The function(s) of these phosphorylations have been difficult to address in detail because of the instability of these modifications because of the inherent catalytic nature of the phosphatase. Possible roles for the phosphotyrosines include the recruitment of phosphotyrosine-binding modules such as SH2 and PTB domain-containing proteins as well as activation of the SHP-1 catalytic activity. One or more of these phosphotyrosines could bind in an intermolecular or intramolecular fashion to the SH2 domains of SHP-1.
Recently, expressed protein ligation (20,21) has been used to probe the function of phosphotyrosine modifications in the related protein-tyrosine phosphatase enzyme, SHP-2 (22). In this study, the phosphotyrosine mimetic phosphonomethylenephenylalanine (Pmp) was incorporated into the C terminus of SHP-2 at two phosphorylation sites and the effects on SHP-2 catalytic behavior determined. Because the SHP-2 and SHP-1 tails show little sequence homology (Fig. 1B), and their SH2 domains appear to show different phosphotyrosine sequence preferences, it was not known whether the results with SHP-2 would extend to SHP-1. Here we use expressed protein ligation to probe the importance of C-terminal tyrosine phosphorylation of SHP-1 and find evidence both for stimulatory and adaptor functions.

EXPERIMENTAL PROCEDURES
Preparation of the Shp-1 Constructs-The full-length human shp-1 tyrosine phosphatase encoded in a DNA plasmid (gift of Dr. Jun Wang) was used to amplify DNA encoding amino acids 1-556 by use of primers containing NdeI and SmaI restriction digestion sites at the 5Ј and 3Ј ends, respectively. The PCR product was purified and ligated into the pTYB2 vector (New England Biolabs) in-frame with the intein and chitin binding domain open reading frames. The extra SmaI site at the 3Ј end of the shp-1 gene was deleted using QuikChange site-directed mutagenesis (Stratagene) to give the plasmid pTYB2-shp-1-(1-556). This plasmid was used to generate 564-modified proteins (after ligation) and as a template to produce all other mutants by the QuikChange method including pTYB2-shp-1-(1-531), which was used to generate 536-modified proteins. All constructs were confirmed by DNA sequencing the entire open reading frames of the shp-1 gene.
Peptide Synthesis-Standard Fmoc-protected amino acids, peptide synthesis reagents, and Wang resins were purchased from Novabiochem. All peptides were synthesized on Wang resin using the standard Fmoc strategy on a Rainin PS-3 machine. Pmp, F 2 Pmp, or phosphotyrosine (Tyr(P)) containing peptides were prepared by incorporating the corresponding non-natural amino acids during the coupling reactions. Crude peptides were purified by reversed-phase high performance liquid chromatography on a preparative or semipreparative C18 column. The purity (Ͼ95%) of the peptides was established by reversed-phase analytical high performance liquid chromatography and the molecular weights were confirmed by both MALDI and electrospray mass spectrometric analysis.
Expressed Protein Ligation-Protein ligations were carried out as described (20,22). Briefly, the shp-1 DNA constructs were transformed into Escherichia coli BL21(DE3) cells and the transformed cells grown at 37°C until A 595 ϭ 0.6 -0.8. The cells were then treated with isopropyl-1-thio-␤-D-galactopyranoside (final concentration ϭ 0.2 mM) and harvested after 20 h culture at 16°C. The resuspended cells were lysed by French press and the lysates were eluted over chitin beads and the beads washed as previously described (20,22). The immobilized SHP-1 fusion proteins were ligated to synthetic peptides in the presence of 2% thiophenol. 2 The purity of the semisynthetic proteins was determined by 10% SDS-PAGE stained with Coomassie Blue and by MALDI mass spectrometry (Ͼ90%). The yield was 2-10 mg of protein/liter in E. coli cell culture as revealed by Bradford assay (24).
Gel Filtration-The molecular weights of the semisynthetic proteins, SHP-1/Tyr 564 , SHP-1/Pmp 564 , SHP-1/Tyr 536 , and SHP-1/Pmp 536 were determined by gel filtration chromatography on a Pharmacia FPLC system. Semisynthetic SHP-1 proteins (100 l of 2 mg/ml) were loaded onto a Superdex 200 column (Amersham Biosciences), which was preequilibrated with 50 ml of buffer (25 mM Na-Hepes, pH 7.2, 200 mM NaCl, and 2 mM dithiothreitol). A molecular weight marker kit containing cytochrome c, carbonic anhydrase, bovine serum albumin, alcohol dehydrogenase, and ␤-amylase proteins (Sigma) was used to obtain a standard curve. The molecular weights of the SHP-1 semisynthetic proteins were calculated from the retention volumes by the use of the standard curve (25).
Phosphatase Assays-pNPP (purchased from Acros) was used as substrate to determine the phosphatase activities of the semisynthetic SHP-1 proteins. It has previously been observed that pNPP shows nonclassical Michaelis-Menten kinetics with SHP-1 (14) and so reaction rates were measured as specific activities using a fixed and subsaturating concentration of pNPP (2 mM). Reactions were carried out typically with 1 M SHP-1 protein and incubated in 50 l of reaction buffer (0.2 mg/ml bovine serum albumin, 100 mM Na-Hepes, pH 7.4, 150 mM NaCl, 1 mM EDTA, and 10 mM dithiothreitol) at 25°C for 30 min. It should be noted that pH 7.4 is above the pH optimum for SHP-1 (14) but was selected because it is likely to be more physiologic. The reactions were quenched with 950 l of 1 N NaOH and the turnover rates of the proteins were calculated from the amount of the released p-nitrophenolate, which was determined from its spectrophotometric absorbance at 405 nm. Reaction rates were shown to be linear with enzyme concentration and time in the ranges used (Fig. 4A). To assess the stimulatory effects of synthetic (phospho)peptides on the phosphatase reactions, 70 M peptide was included in the reaction buffer prior to initiation with SHP-1.
Grb2 Binding Studies-GST-Grb2 fusion protein was expressed in E. coli BL21(DE3) cells and purified by the use of glutathione-agarose (Sigma) as previously described (26,27). The Grb2 binding studies were carried out as follows: 50 l of GST-Grb2 bead slurry containing ϳ50 g of fusion protein (estimated by 10% SDS-PAGE), was centrifuged at 2000 ϫ g for 5 min in a small plastic (Eppendorf) tube. The supernatant was discarded and the pellet was washed twice with 0.1 ml of binding buffer (25 mM Na-Hepes, pH 7.5, 10% glycerol, and 100 mM NaCl), and then 10 l of binding buffer was added to the pellet to resuspend the beads. To the resuspended beads was added 10 g of SHP-1 protein in 30 l of binding buffer. After 4 min incubation at 16°C, the reaction mixture was centrifuged at 2000 ϫ g for 5 min. The pellets were then washed three times with 200 l of binding buffer. The beads were finally mixed with 30 l of 1ϫ SDS loading dye at 95°C for 5 min and the supernatant was loaded onto 10% SDS-PAGE. The dried gels were scanned and the amount of SHP-1 proteins bound to Grb2 was quantified using ImageQuant software.

RESULTS
Generation of Semisynthetic SHP-1-The N-terminal portion of SHP-1 including its SH2 domains and catalytic domain but lacking most of the C-terminal tail was subcloned into the pTYB2 vector in-frame with the intein-chitin binding domains. Two constructs were developed covering residues 1-531 and 1-556, which enable the ligation of 15-and 16-residue Nterminal cysteine containing peptides, respectively, to replace the physiologic sequences in these positions (Fig. 1C). Based on previously published studies, it was expected that partial deletions in the C-terminal sequence would resemble full-length protein catalytic behavior (14,15), and this proved to be the case in our hands (data not shown). The requisite synthetic peptides containing either tyrosine or Pmp at the sites of phosphorylation were prepared by solid phase peptide synthesis. Cysteines were installed at the N terminus of these peptides replacing Gln (for 564-modified proteins) or Ser (for 536-modified proteins) in the natural sequences as required for the expressed protein ligation method. To facilitate ligation, the C-terminal residues of the recombinant protein were Gly, which was the natural residue of the 536-modified proteins but replaced a Ser in the 564-modified proteins. These mutations did not significantly affect the activities of the proteins as revealed by pNPP assay (data not shown). Expressed protein ligation performed in the presence of thiophenol 2 as previously described (20,22) led to robust protein production (Ͼ2 mg/liter of E. coli cell culture), which appeared to be greater than 90% pure eluting from the chitin resin ( Fig. 2A). Mass spectra supported the efficiency of the ligation process (Fig. 2B). It was also demonstrated that the phosphatase activity of standard SHP-1 recombinant protein identical in sequence length to the semisynthetic proteins showed similar catalytic behavior (data not shown).
Gel Filtration-There was a theoretical possibility that tail phosphorylation could promote intermolecular protein interactions via SH2 domain-phosphotyrosine interactions. This potential was assessed with the semisynthetic proteins by size exclusion chromatography (Fig. 3). A comparison of elution profiles of Pmp-containing versus tyrosine containing proteins showed no significant differences in the elution profiles (Fig. 3). Moreover, by comparison to the profiles of protein standards (25) it was shown that each of the semisynthetic proteins showed sizes consistent with monomers. Thus it can be concluded that at the concentration investigated (ϳ1-2 mg/ml), the Pmp groups do not promote protein dimerization or higher order oligomers.
Enzyme Activity of Pmp-substituted Semisynthetic SHP-1 Proteins-To address the possibility that the Pmp substitutions could modulate the enzyme activity of SHP-1, the relative activities of the semisynthetic proteins were studied with the well established substrate, pNPP. Because wild-type SHP-1 does not display Michaelis-Menten kinetic behavior, a fixed concentration of pNPP was employed. As can be seen, SHP-1/ Pmp 536 showed a substantial 4-fold increase in phosphatase activity compared with SHP-1/Tyr 536 (Fig. 4). In contrast, SHP-1/Pmp 564 showed very similar catalytic behavior compared with SHP-1/Tyr 564 . These results suggest site-selective activation of SHP-1 by tyrosine phosphorylation at Tyr 536 (Fig. 4). Analogous results were seen using phosphorylated RCM-lysozyme as substrate (data not shown).
To probe the structural basis of Pmp 536 activation, the phosphatase activity of the semisynthetic proteins was measured in the presence of phosphotyrosine peptide EpoR-pY-429 (AcHN-PHLKYLpYLVVSDK-CO 2 H). This peptide has been shown to stimulate the phosphatase activity of SHP-1 by relieving basal inhibition mediated by the N-terminal SH2 domain (15). As can be seen (Fig. 4B), 70 M EpoR-pY-429 caused a similar 4-fold stimulation of SHP-1/Tyr 564 and SHP-1/Pmp 564 . In contrast, whereas EpoR-pY-429 led to a 4-fold activation of SHP-1/ Tyr 536 , it caused a more modest 1.5-fold activation of SHP-1/ Pmp 536 . This is consistent with the possibility that Pmp 536 can partially relieve the basal suppression mediated by the Nterminal SH2 domain.
Generation of a Semisynthetic SHP-1 with a Consensus Sequence-Because Pmp 536 may be interacting with the N-termi-FIG. 1. Scheme of the SHP-1 structure and the strategies for modification. A, SHP-1 protein consists of two SH2 domains followed by one catalytic domain (PTPase) and a C-terminal tail. R30 and R136 are conserved Arg residues within the FLVRES motifs thought to be important for the N-SH2 and C-SH2 domains, respectively, for interaction with phosphotyrosine containing peptides. C453 (Cys 453 ) is critical for the activity of the SHP-1 protein. Two tyrosine residues at the C-terminal tail, Tyr 536 and Tyr 564 , can be phosphorylated by tyrosine kinases; B, comparison of the C-terminal tail sequences of the SHP-1 and SHP-2 proteins. Tyrosine residues that can be phosphorylated are indicated; C, sequences of the C-terminal tails of the semisynthetic proteins. The sequences of the synthetic peptides used in expressed protein ligation are underlined. To facilitate the protein ligation, Gln 532 (536-modified SHP-1), Ser 556 (564-modified SHP-1), and Ser 557 (564-modified SHP-1) were mutated to Cys 532 , Gly 556 , and Cys 557 , respectively. The mutated residues are depicted in bigger font. X represents tyrosine or phosphotyrosine analogues, either Pmp or F 2 Pmp; D, the structures of tyrosine, Pmp, and F 2 Pmp are shown. nal SH2 domain of SHP-1 to enhance the phosphatase activity of SHP1, we considered the possibility that an amino acid sequence surrounding Pmp 536 , which was designed to interact with the N-SH2 domain, might show more robust stimulation of SHP-1. Based on the analysis of the affinity of the SHP-1 N-terminal SH2 domain using peptide libraries (28), a peptide containing the consensus sequence LHpYMNF (Fig. 5A) was synthesized for use in expressed protein ligation. This semisynthetic protein (SHP-1/Pmp 536con ) was prepared as efficiently as SHP-1/Pmp 536 and subjected to catalytic activity measurements. As shown in Fig. 5B, the activity of SHP-1/ Pmp 536con was greater than SHP-1/Pmp 536 and minimally affected by the addition of EpoR-pY-429. This provides further evidence of an interaction between Tyr(P) 536 and the N-terminal SH2 domain. (29) that F 2 Pmp is a closer mimic of phosphotyrosine compared with Pmp. Because the consensus sequence for N-SH2 binding led to greater SHP-1 activation relative to the natural sequence, we pursued incorporation of F 2 Pmp at residues 536 and 564 of SHP-1, respectively. Synthesis of the difluoro analogue was carried out as described by Guo et al. (23) and incorporated into the same synthetic peptide sequences used with Pmp (Fig. 1C). These peptides were used to produce the desired semisynthetic proteins in similar efficiency compared with the Pmp derivatives. Strikingly, F 2 Pmp induced greater phosphatase activity at both positions 536 and 564 (Fig. 5C). An 8-fold activation in phosphatase activity was detected for SHP-1/F 2 Pmp 536 compared with SHP-1/Tyr 536 , and SHP-1/F 2 Pmp 536 did not show enhanced activity in the presence of EpoR-pY-429. Moreover, SHP-1/F 2 Pmp 564 actually displayed a 1.6-fold increase in activity compared with SHP-1/Tyr 564 or SHP-1/Pmp 564 . These results unmask a potential for tyrosine phosphorylation to provide modest stimulation of SHP-1 at the 564 position. This modest stimulation in the absence of peptide EpoR-pY-429 was less apparent in the presence of stimulatory peptide (Fig. 5C) sibly involves intramolecular interactions between the phosphonates and the SH2 domains. The molecular basis of SH2phosphotyrosine recognition generally involves a highly conserved SH2 Arg (within the conserved motif FLVRES) forming an electrostatic interaction with the phosphate moiety of the ligand (30). It is typically the case that mutation of this Arg to other residues (including Lys) will greatly weaken the interaction with ligand but otherwise be structurally tolerated. We changed this key residue (Arg 30 ) within the N-SH2 domain of SHP-1 to several other residues including Ala, Met, Ile, Leu, and Lys in SHP-1 semisynthetic proteins. These semisynthetic mutant proteins were prepared identically to the "wild-type" versions described above. In each case, in the absence of phosphotyrosine ligands significant increases (2-6-fold) in the basal activity of the SHP-1 PTPase were observed (see Fig. 6A). This unusual finding confirms and extends an earlier observation of Pei et al. (15) and suggests that Arg 30 is directly or indirectly contributing to the presumed intramolecular inhibitory interaction between the N-SH2 domain and the catalytic domain. In contrast, mutation of the homologous residue in the C-SH2 domain (Arg 136 ) does not affect basal SHP-1 activity (see Fig. 6A).

Effect of F 2 Pmp Incorporation on SHP-1-It has previously been shown by Burke and colleagues
Perhaps more surprisingly, the phosphopeptide EpoR-pY-429 was able to further stimulate each of these Arg 30 mutant SHP-1 enzymes at a concentration identical to that used in the wild-type SHP-1 studies (Fig. 6A). Using the corresponding nonphosphorylated form of the peptide (EpoR-Y-429) at the same concentration, no activation of the Arg 30 mutants was detected (data not shown). Thus the putative interaction of phosphopeptide ligand with the N-SH2 domain of SHP-1 is unconventional compared with typical cases and the role of the FLVRES Arg is likely substituted by another residue in this case.
The effects of mutation of Arg 30 on several of the phosphonate containing semisynthetic proteins were also examined (Fig. 6, B and C). It can be seen that SHP-1/R30A-Pmp 536 shows a 2-fold greater PTPase rate compared with SHP-1/ R30A-Tyr 536 (Fig. 6B). These results are consistent with the intermolecular studies with phosphopeptide EpoR-pY-429 in which it was shown that the presence of Arg 30 is not critical for PTPase stimulation. It was also shown that SHP-1/R136A-Pmp 536 displays a 4-fold greater rate compared with SHP-1/ R136A-Tyr 536 arguing against the role of the C-SH2 domain in mediating PTPase activation by the Pmp 536 modification. Finally, it was found that SHP-1/R136A-F 2 Pmp 564 shows an identical rate to SHP-1/R136A-Tyr 564 (Fig. 6C) suggesting that the C-SH2 domain is most likely responsible for the modest activation by the F 2 Pmp 564 modification (Fig. 6C). Taken together, the data support a structural model of SHP-1 activation by phosphorylation shown in Fig. 7.
Grb2 Binding-The role of the phosphonate modifications of SHP-1 in mediating interactions with Grb2, an SH2 domain containing adaptor protein (22,26), were studied by a variant of the GST pull-down assays. GST-Grb2 was immobilized on glutathione-agarose and incubated with SHP-1 semisynthetic proteins for 4 min prior to several brief buffer washes. It was found that both SHP-1/Pmp 536 and SHP-1/Pmp 564 proteins showed significantly greater binding than the unphosphonylated semisynthetic proteins (Fig. 8). Likewise, F 2 Pmp SHP-1 proteins showed greater binding to Grb2 than unmodified proteins. Interestingly, Pmp containing SHP-1 proteins showed somewhat more efficient binding (about 2-fold) to Grb2 compared with F 2 Pmp proteins. This could be the case because F 2 Pmp have a lower relative affinity for Grb2, a slower rate of release from their intramolecular interactions with the SHP-1 SH2 domains, or a slower on-rate with respect to the Grb2 domains (these assays were designed to look at kinetic effects). However, the importance of slow intramolecular dissociation is argued against because mutations in the N-SH2 or C-SH2 domains did not have a large effect on the relative proportion of SHP-1 proteins bound to Grb2. DISCUSSION The value of expressed protein ligation in addressing structure-function relationships in reversible phosphorylation of signaling proteins is becoming increasingly well demonstrated (20 -22, 31, 32). It is most conveniently applied when the region of the protein to be modified is in the C terminus of a signaling protein because the C-terminal thioester generated using intein technology can be fused with a readily prepared N-terminal cysteine containing peptide. Recently published studies on transforming growth factor-␤-substrate interactions (31), Csk-Src regulation (20,32), and SHP-2 tyrosine phosphatase (22) employing expressed protein ligation have illustrated the power and scope of the method. The recent work on SHP-2 illustrates the strength of the use of nonhydrolyzable phosphotyrosines to examine the structure and function of a signaling protein (22). Whereas for many years Glu and Asp have been used as mimics of phosphoserine and phosphothreonines, often successfully, no such encoded amino acid comes close to resembling phosphotyrosine. Thiophosphates, resistant to enzymatic hydrolysis, have been introduced into signaling proteins by enzymatic modification of recombinant proteins with ATP␥S. These are often sluggish reactions that generally afford low stoichiometries and often show poor regioselectivity (33). Moreover, thiophosphates can indeed be hydrolyzed chemically and enzymatically, by some phosphatases almost as rapidly as phosphate substrates (34), in contrast to the nonhydrolyzable analogues used here.
The poor homology of SHP-2 and SHP-1 C-terminal tail regions (Fig. 1B) made it difficult to predict based on the results of SHP-2 what to expect with SHP-1. It is now clear that phosphorylation at Tyr 536 can activate the SHP-1 PTPase activity in analogy with Tyr 542 in SHP-2. However, the degree of activation, about 8-fold with F 2 Pmp in SHP-1, is significantly greater than the 2-3-fold observed with Pmp and SHP-2. It remains to be seen if this difference is related to the choice of the phosphotyrosine analogue although given that only a 4-fold effect was seen with Pmp 536 and SHP-1, this has to be considered a possibility. It should also be pointed out that a study with enzymatically phosphorylated SHP-1 at the 536 position also reports PTPase activation by this modification (18), although this was characterized with a less pure protein mixture and uncertain stoichiometry in which the SHP-1 undergoes dephosphorylation during the PTPase measurements.
The model of activation based on the findings with the consensus sequence engineered SHP-1, stimulation with EpoR-pY-429 peptide, and mutagenesis support the proposition that phosphoryl modification at 536 activates SHP-1 by interaction with the N-terminal SH2 domain (Fig. 7). Whereas mutation of the FLVRES Arg of the N-SH2 leads to somewhat unconventional behavior, it is clear that mutation of the C-SH2 domain does not play a role in activation by Pmp 536 (Fig. 7). The gel filtration experiments showing that SHP-1/Pmp 536 is monomeric support the likelihood that this interaction is intramolecular (Fig. 7). Furthermore, this activation model would be similar to the effect of phosphorylation of Tyr 542 in SHP-2 (22). Not addressed in these studies or in Fig. 7 is the interesting finding that deletion of the C-terminal tail 35 amino acids but not the final 60 amino acids appears to be activating for SHP-1 (15). The basis of catalytic activation by this truncation is not yet understood and may add complexity to the simple model in Fig. 7. It should be stated that deletion of the C-terminal 49 amino acid residues as present in SHP-1/Tyr 536 or 23 amino acids in SHP-1/Tyr 564 showed similar catalytic activity to fulllength SHP-1 in our hands.
It is worthwhile to consider the reason why mutation of the FLVRES Arg itself leads to PTPase stimulation and yet these mutant proteins are still susceptible to phosphotyrosine activation. One possibility is that the SHP-1 N-SH2 domain is noncanonical and the FLVRES Arg contributes relatively little to binding in this peculiar case. Another explanation would be that the SHP-1 catalytic domain is perturbing the structure of the N-SH2 domain such that it behaves anomalously. Related to this idea, the catalytic domain could even be contributing residues that substitute for SH2 domain residues in the function of the domain. Finally, it should be mentioned that a loss of affinity of the N-SH2 domain for the catalytic domain may occur when Arg 30 is mutated, leading to PTPase activation. This weaker interaction may make the N-SH2 domain, even though damaged by Arg 30 mutation, more available for phosphotyrosine binding, which would offset the loss of affinity resulting from the Arg 30 -phosphate interaction. Further structural studies will be needed to distinguish among these possibilities.
In contrast to the relatively large stimulation by phosphonates at the 536 position of SHP-1, the catalytic stimulation from F 2 Pmp at the 564 site is rather modest at 1.6-fold. It is noteworthy that the effect seen with F 2 Pmp was absent when the initial Pmp analogue was employed. This argues for the utility of the fluorophosphonate analogues bringing out more subtle structural effects because they are more faithful mimics. Whereas a small effect, the F 2 Pmp effect can be structurally A, relative level of Pmp 536 -modified SHP-1 semisynthetic proteins carrying SH2 mutations binding to Grb2; B, comparison of Pmp-and F 2 Pmp-modified SHP-1 semisynthetic proteins binding to Grb2; C, relative levels of SHP-1/F 2 Pmp 564 carrying SH2 mutations binding to Grb2. GST-Grb2 was immobilized on glutathione resin and the level of SHP-1 protein "pulled-down" was measured as described under "Experimental Procedures." One SDS-PAGE, stained with Coomassie Blue, of three independent experiments, with good reproducibility, is shown. The mean relative intensities from the three experiments are indicated below each lane. rationalized because it disappears upon mutation of the C-SH2 Arg 136 residue. Thus, it is likely that this phosphorylation of Tyr 564 can modestly enhance SHP-1 activity by intramolecular C-SH2 domain engagement, presumably by an indirect effect (Fig. 7). This is qualitatively similar to the behavior of Tyr 580 in SHP-2 although in that case the activation is more robust and observable with Pmp (2-3-fold), comparable with that of Tyr 542 phosphorylation in SHP-2 (22).
Interestingly, although phosphonylation of SHP-1 at 536 and 564 can allow for intramolecular interactions with the SHP-1 SH2 domains, they are still quite efficiently able to facilitate binding to the adaptor protein Grb2 (Fig. 8). These results should be contrasted with Tyr 542 phosphorylation of SHP-2 where intramolecular engagement with the N-SH2 domain effectively competes with Grb2 binding kinetically. Thus, tail phosphorylation of SHP-1 could presumably impact signal transduction in vivo by either catalytic activation or by an adaptor function. The particular pathway and interacting molecules may dictate which effects predominate in the cellular environment.
Summary-Expressed protein ligation has been used to generate site-specific and stoichiometric phosphonate-modified forms of SHP-1 to simulate the effects of tyrosine phosphorylation of SHP-1. These studies describe the first comparative analysis of the effects of difluoromethylenephosphonate and methylenephosphonate as phosphotyrosine mimetics site-specifically incorporated within the context of a protein. It was shown that phosphonate at the 536 position of SHP-1 is capable of up to 8-fold stimulation of the SHP-1 tyrosine phosphatase activity, likely by intramolecular engagement of the N-SH2 domain, relieving basal inhibition. In contrast, phosphonate modification of the 564 position of SHP-1 results in a smaller (1.6-fold) stimulation of the tyrosine phosphatase activity, probably by an indirect effect via interaction with the C-terminal SH2 domain. The phosphonate-modified SHP-1 proteins are readily able to recruit the SH2-containing adaptor protein Grb2 suggesting that the intramolecular interactions promoted by tyrosine phosphorylation are not highly favorable energetically. These studies suggest that tyrosine phosphorylation of SHP-1 could play distinct roles in cell signaling either by direct catalytic activation of the enzyme or by recruitment of other signaling molecules to a specific cellular location.