SHP1 and SHP2 Protein-tyrosine Phosphatases Associate with βc after Interleukin-3-induced Receptor Tyrosine Phosphorylation

The cytoplasmic tyrosine phosphatases, SHP1 and SHP2, are implicated in the control of cellular proliferation and survival. Here we demonstrate that both SHP1 and SHP2 associate with the βc subunit of the human interleukin-3 (IL-3) receptor following IL-3 stimulation and that the src homology region 2 (SH2) domains of these phosphatases mediate this interaction. Sequential immunoprecipitation analyses suggest this interaction is direct. Competition studies, using phosphotyrosine-containing peptides based on sequences surrounding key tyrosine residues within βc, suggest that phosphorylation of tyrosine 612 is the key event mediating the association of βc with SHP1 and SHP2. However, inhibition of SHP2 binding to βc, did not prevent tyrosine phosphorylation of SHP2. Interestingly, this same phosphopeptide served as a substrate for the tyrosine phosphatase activity of both SHP1 and SHP2. Binding of these protein-tyrosine phosphatases to the IL-3 receptor may regulate IL-3 signal transduction pathways, both through their catalytic activity and through the recruitment of other molecules to the receptor complex.

The cytoplasmic tyrosine phosphatases, SHP1 and SHP2, are implicated in the control of cellular proliferation and survival. Here we demonstrate that both SHP1 and SHP2 associate with the ␤c subunit of the human interleukin-3 (IL-3) receptor following IL-3 stimulation and that the src homology region 2 (SH2) domains of these phosphatases mediate this interaction. Sequential immunoprecipitation analyses suggest this interaction is direct. Competition studies, using phosphotyrosinecontaining peptides based on sequences surrounding key tyrosine residues within ␤c, suggest that phosphorylation of tyrosine 612 is the key event mediating the association of ␤c with SHP1 and SHP2. However, inhibition of SHP2 binding to ␤c, did not prevent tyrosine phosphorylation of SHP2. Interestingly, this same phosphopeptide served as a substrate for the tyrosine phosphatase activity of both SHP1 and SHP2. Binding of these protein-tyrosine phosphatases to the IL-3 receptor may regulate IL-3 signal transduction pathways, both through their catalytic activity and through the recruitment of other molecules to the receptor complex.
Interleukin-3 (IL-3) 1 stimulates the growth, survival, and differentiation of a broad range of hemopoietic cells, including pluripotent stem cells and progenitors, mast cells, megakaryocytes, macrophages, neutrophils, and basophils (1). The high affinity human IL-3 receptor is composed of a specific ␣ subunit of 60 -70 kDa and a ␤ subunit of 130 -140 kDa (2), which is also a component of the high affinity human GM-CSF and IL-5 receptors (3,4) and is referred to as ␤c. Although lacking intrinsic tyrosine kinase activity, upon IL-3 binding, ␤c be-comes phosphorylated on multiple tyrosine residues, some of which serve as selective binding sites for molecules containing SH2 domains. A number of cellular proteins including JAK-2 (5), , SHP2 (7), Shc (8), Erk-1 and Erk-2 (9) are also inducibly tyrosine-phosphorylated in response to IL-3 stimulation.
A subgroup of cytoplasmic protein-tyrosine phosphatases (PTPases), characterized by containing two SH2 domains NH 2terminal to their catalytic phosphatase domain, are now referred to as SHP, for SH2 domain containing phosphatases. SHP1, also referred to as PTP-1C (10), HCP (11), SH-PTP1 (12), and SHP (13), is expressed almost exclusively in lymphohemopoietic cells, but has also been detected in epithelial cells (12). Mutations within SHP1 are responsible for the phenotype of the two moth-eaten mouse strains (me/me and me v /me v ; Refs. 14 and 15). SHP1 appears to act as a negative regulator of growth factor signaling, as these moth-eaten mice die soon after birth due to overproliferation and accumulation of macrophages in the lungs (16). In addition, increased SHP1 levels have been shown to supress cell growth in response to IL-3 (17). Activation of certain hemopoietic growth factor receptors by their ligands results in the association of SHP1 with the EpoR, via tyrosine 429 (18,19), c-kit (20), and the murine IL-3 receptor ␤ subunit, Aic2A (17), the sites of interaction with which are not currently known.
We have shown previously that treatment of responsive cells with IL-3 or GM-CSF results in the tyrosine phosphorylation of SHP2, creating a docking site (Tyr 304 or Tyr 542 ) for the SH2 * This work was supported by core funding from The School of Pharmacy and Pharmacology (to M. J. W.), the Medical Research Council (Canada) (to J. W. S.), and the Canadian Arthritis Society (to F. J.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶  1 The abbreviations used are: IL, interleukin; ␤c, human IL-3 receptor ␤ common subunit; Epo, erythropoietin; GST, glutathione S-transferase; EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; PDGF, platelet-derived growth factor; PDGFR, platelet-derived growth factor receptor; PTPase, protein-tyrosine phosphatase; pY, phosphotyrosine; SCF, stem cell factor; SH2, src homology region 2; GM-CSF, granulocyte macrophage-colony stimulating factor; Bistris, domain of Grb2 (7). In addition, we found that IL-3 treatment of cells resulted in co-precipitation of phosphatidylinositol 3Јkinase with SHP2, as well as increasing the phosphatase activity of SHP2 (7). We were interested in determining other interactions mediated by SHP2 and SHP1 in response to IL-3, to investigate the possible roles for these PTPases in hemopoietic cells. Here we provide evidence that in vitro and in vivo, both SHP1 and SHP2 bind directly to the phosphorylated form of the IL-3 receptor. Using synthetic phosphopeptides we have identified potential sites for this binding and shown that these peptides can also be utilized as substrates by the SHP1 and SHP2 catalytic domains in in vitro assays.
GST Fusion Proteins-The construction, expression, and purification of SHP1SH2-GST and SHP2SH2-GST fusion proteins have been described previously (42).
Phosphopeptide Design and Synthesis-The synthesis, purification, and mass spectroscopic analysis of phosphotyrosine-containing peptides has been described (43,44). Table I lists the phosphopeptides used in this study. Throughout the text the phosphotyrosine-containing peptides are referred to by the relative position of the tyrosine residue in ␤c, lacking the 14-amino acid signal peptide.
Phosphopeptide Competition Analyses-SHP1SH2-GST and SHP2SH2-GST fusion proteins were preincubated with the phosphopeptides at 100 M or 50 M and glutathione-Sepharose by rotating at 4°C for 60 min. Cell extracts, from 1 ϫ 10 7 cells, also containing the appropriate concentration of phosphopeptide, were added to the preincubated mixture and rotated at 4°C for a further 60 min. For the anti-SHP2 phosphopeptide competition analyses, 1 ϫ 10 7 cells were lysed in the presence of 500 M phosphopeptide 612 or 750 or 100 M 1009 and immunoprecipitated with the anti-SHP2 antibody as described above.
SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting-SDS-polyacrylamide gel electrophoresis and immunoblotting were carried out as described previously (8,45). Primary antibodies were used at 0.1 g/ml for the monoclonal anti-phosphotyrosine antibody 4G10 and the polyclonal antibodies raised against ␤c, SHP1 and SHP2 were used at 0.5 g/ml. Goat anti-mouse or anti-rabbit horseradish peroxidaseconjugated secondary antibodies (Dako) were used at 0.1 g/ml. Immunoblots were developed using the ECL system (Amersham Corp.) and Kodak X-AR5 film. Blots were stripped as described previously (45).
Phosphatase Assays-SHP2 and SHP1 GST fusion proteins were expressed, purified, and thrombin-cleaved as described previously (42,43). To assay the dephosphorylation of the ␤c subunit phosphopeptides, the release of P i was measured by a Malachite Green assay in halfvolume 96-well plates (Costar) at 23°C (44). SHP1(⌬SH2/⌬CT; Ref. 42) assays were performed in 25 mM Bistris-propane/HCl, pH 6.5, 50 mM NaCl, and 10 mM 2-mercaptoethanol. The assay buffer for SHP2(⌬SH2; Ref. 42) was the same except for 25 mM Bistris-propane/HCl, pH 6.25. Reaction mixtures containing assay buffer, 1 mM phosphopeptides, and 45 nM SHP1(⌬SH2/⌬CT) or 20 nM SHP2(⌬SH2) were incubated at room temperature with occasional agitation. After several time points, 25-l samples were removed and added to 50 l of Malachite Green reagent to detect the release of inorganic phosphate. The absorbance at 620 nm was measured and compared with a standard P i curve.

RESULTS
Tyrosine-phosphorylated ␤c Co-precipitates with SHP1 and SHP2 after IL-3 Stimulation-It has been demonstrated previously that SHP1 can associate with the murine IL-3 receptor ␤ subunit, Aic2A, via an unmapped site (17). No such association has been demonstrated in human cells for SHP1 or SHP2. However, we had observed previously that SHP2 was inducibly tyrosine-phosphorylated in response to IL-3, at a site that enabled binding of Grb2. In addition we also observed coprecipitation of SHP2 with phosphatidylinositol 3Ј-kinase and activation of SHP2, raising the possibility that SHP2 has multiple functions in IL-3 signaling (7). Therefore, we were interested in determining whether SHP1 and SHP2 were able to act as adaptor molecules by binding to the human IL-3 receptor ␤ subunit (␤c). TF-1 cells were left untreated as a control or stimulated with IL-3, lysed, and immunoprecipitates prepared with either monoclonal anti-␤c or polyclonal anti-SHP1 or anti-SHP2 antibodies. Immunoblotting was performed on the precipitates with the 4G10 anti-phosphotyrosine antibody, the results are shown in Fig. 1.
A tyrosine-phosphorylated 130-kDa protein was immunoprecipitated with the anti-␤c monoclonal antibody after IL-3 stimulation (Fig. 1A, lane 4). Reprobing this same blot with a polyclonal anti-␤c antibody confirmed this to be ␤c (Fig. 1B,  lanes 3 and 4). The anti-SHP1 antibodies precipitated three tyrosine-phosphorylated proteins of 130, 105, and 60 kDa (Fig.  1A, lanes 5 and 6). The 60-kDa phosphoprotein was present in SHP1 precipitates from both control and IL-3-stimulated cells (Fig. 1A, lanes 5 and 6), and blotting with anti-SHP1 antibodies (Fig. 1C, lanes 5 and 6) confirmed that this protein was SHP1. SHP1 is constitutively phosphorylated in other hemopoietic cells (17), and this also appears to be the case in TF-1 cells. The 130-and 105-kDa species were only observed in the SHP1 precipitates from IL-3-treated cells. The identity of the 105-kDa protein is unknown; however, the 130-kDa species comigrated with the tyrosine-phosphorylated ␤c precipitated by the anti-␤c antibody (Fig. 1A, lane 4), suggesting that tyrosinephosphorylated ␤c co-precipitates with SHP1. We could not confirm this by reprobing this blot with polyclonal anti-␤c antibodies (Fig. 1B, lane 6), probably because the amount of tyrosine-phosphorylated 130-kDa species precipitated by the anti-SHP1 antibodies (Fig. 1A, lane 6) was considerably less than that precipitated by the anti-␤c antibodies (Fig. 1A, lane  4) and so below the limits of detection.
The SHP2 antibodies precipitated four tyrosine-phosphorylated proteins of 70, 72, 90, and 130 kDa from IL-3-stimulated TF-1 cells (Fig. 1A, lane 8). The doublet at 70 and 72 kDa corresponds to SHP2, as confirmed by immunoblotting with the SHP2 antibody (Fig. 1D, lanes 7 and 8). The identity of the 90-kDa protein is unknown. The 130-kDa protein co-migrated with tyrosine-phosphorylated ␤c precipitated by the anti-␤c antibodies (Fig. 1A, lane 4), and its identity as ␤c was confirmed by reprobing the blot with the polyclonal anti-␤c antibodies (Fig. 1B, lane 8). In reciprocal experiments, in which blots of material precipitated with anti-␤c were reprobed with antibodies specific for either SHP1 or SHP2, neither were detectable (Fig. 1, C and D). The monoclonal anti-␤c antibody precipitates less than 10% of the ␤c expressed in the TF-1 cells, 2 only a proportion of which is likely to be phosphorylated at the appropriate sites and will hence interact with downstream signaling molecules. Thus, our failure to detect SHP1 and SHP2 in the anti-␤c precipitates most likely reflects the fact that the amounts present were below our limits of detection.
SHP1 and SHP2 Bind ␤c through Their SH2 Domains after IL-3 Stimulation-We next investigated whether the SH2 domains of the two PTPases mediated the observed interactions with ␤c. Precipitates from extracts of TF-1 were prepared using fusion proteins in which GST was fused to either the SH2 domains of SHP1 (SHP1SH2-GST) or SHP2 (SHP2SH2-GST). Both SHP1SH2-GST and SHP2SH2-GST precipitated a tyrosine-phosphorylated 130-kDa protein from extracts of cells stimulated with IL-3 ( Fig. 2A, lanes 2 and 4) but not from control samples ( Fig. 2A, lanes 1 and 3), which was recognized by the polyclonal anti-␤c antibody ( Fig. 2A, lower panel). These results clearly demonstrated that the SH2 domains of SHP1 and SHP2 can associate in vitro with the tyrosine-phosphorylated ␤c following IL-3 stimulation. The additional proteins observed in the antibody immunoprecipitations were not observed in the SH2 domain-fusion protein precipitates. This could be for a number of reasons: (i) they interact with regions of the phosphatase other than the SH2 domains, (ii) they bind the tyrosine-phosphorylated phosphatases themselves, or (iii) they are bound to other proteins in the co-immunopreciptated complex.
To ascertain whether the interactions between ␤c and the SH2 domains of SHP1 and SHP2 were direct, we performed sequential immunoprecipitations. Extracts of TF-1 cells were immunoprecipitated with the monoclonal anti-␤c antibody and the precipitated material boiled and denatured in the presence of SDS and 2-mercaptoethanol. The denatured primary immunoprecipitates were diluted 1 in 10 (so the concentration of SDS was Ͻ0.1%) and re-precipitated with either SHP1SH2-GST or SHP2SH2-GST. The primary anti-␤c immunoprecipitation precipitated tyrosine-phosphorylated ␤c from cells treated with IL-3 (Fig. 2B, lane 2), but not from control cells (Fig. 2B, lane 1). The secondary precipitations with SHP1SH2-GST and SHP2SH2-GST re-precipitated tyrosine-phosphorylated ␤c from the IL-3-stimulated cell extracts (Fig. 2B, lanes 4 and 6). Similar results were observed when the fusion proteins were used as the primary precipitants and anti-␤c as the secondary precipitating agent (data not shown). These results demonstrate that the SH2 domains of SHP1 and SHP2 can bind directly to tyrosine-phosphorylated ␤c.
Binding of SHP1 and SHP2 to ␤c Is Inhibited by a Phosphotyrosine-containing Peptide Based on Tyrosine 612 of ␤c-To confirm that the observed associations involved interactions of the SH2 domains with phosphotyrosine and to provide an indication as to which of the potential tyrosines on ␤c was responsible for mediating the interactions with the SH2 domains of SHP1 and SHP2, we used a peptide competition assay. The tyrosine residues within the ␤c that become phosphorylated upon IL-3 stimulation have not been biochemically mapped, so phosphopeptides corresponding to the sequences surrounding 2 M. J. Welham, unpublished data.  1 and 2). B, cell extracts from the equivalent of 4 ϫ 10 7 cells/ sample were precipiated with 20 g of monoclonal anti-␤c antibody (␣-␤c). Primary ␣-␤c precipitates (1°IP) were eluted and denatured by boiling in SDS sample buffer and 1 ⁄10 of the sample reserved for the primary immunoprecipitation sample (lanes 1 and 2). Secondary precipitations (2°IP) were prepared with either the SHP1SH2-GST (lanes 5 and 6) or SHP2SH2-GST (lanes 3 and 4). Immunoblotting was performed first with 4G10 anti-phosphotyrosine antibodies (␣-PY, A and B, upper panels). The same immunoblot as in A was stripped and reprobed with polyclonal anti-␤c antibodies. The positions of the molecular mass standards are shown and expressed in kilodaltons, and the position of ␤c is indicated. five of the tyrosine residues within ␤c (see Table I) were tested for their ability to block precipitation of the tyrosine-phosphorylated ␤c by SHP1SH2-GST and SHP2SH2-GST. Phosphopeptides corresponding to residues surrounding tyrosines 806 and 856 were not tested, as mutants of ␤c that were truncated at residue 763 or beyond retain normal functions in response to IL-3 (46,47). The results of the competition analyses are shown in Fig. 3, A and B. The phosphopeptide incorporating tyrosine 612 (612) inhibited the precipitation of tyrosine-phosphorylated ␤c by SHP1SH2-GST (Fig. 3A, lane 7). Some inhibition was also consistently observed with phosphopeptide 750 (Fig.  3A, lane 5), but not consistently with the other phosphopeptides. Only phosphopeptide 612 consistently inhibited the precipitation of tyrosine-phosphorylated ␤c by SHP2SH2-GST (Fig. 3B, lane 4). Reprobing of this same blot with polyclonal anti-␤c antibodies confirmed that the presence of phosphopeptide 612 had inhibited precipitation of ␤c by SHP2SH2-GST (Fig. 3B, lower panel). SHP1 and SHP2 each contain two SH2 domains, which appear to differ in their functions (48). Therefore, we used combinations of phosphopeptide 612 with the other phosphopeptides to examine the possibility that any of the latter might make a secondary contribution to the binding of SHP2 to ␤c. At a concentration of 100 M (Fig. 3C, lane 3) phosphopeptide 612 almost completely inhibited (Ͼ95%) precipitation of tyrosinephosphorylated ␤c by SHP2SH2-GST. At 50 M 612, inhibition was approximately 80% (Fig. 3C, lane 4). Combinations of the other phosphopeptides (50 M) with 612 (50 M) did not result in further reduction of the amount of tyrosine-phosphorylated ␤c precipitated by SHP2SH2-GST (Fig. 3C, lanes 5-8). Therefore, it appears that the SH2 domains of SHP2 interact solely with residues surrounding phosphotyrosine 612 of ␤c. Technical difficulties have prevented clearcut results from being ob-tained in similar experiments with SHP1.
We also performed phosphopeptide competition analyses of anti-SHP2 immunoprecipitates from control and IL-3-treated TF-1 cells to investigate whether endogenous SHP2 could also be inhibited from binding to the tyrosine-phosphorylated ␤c by 612. A phosphopeptide incorporating the residues surrounding tyrosine 1009 within the PDGFR (1009; Ref. 42) has been shown to be the binding site for SHP2 (27,31) and was used as a control. Cells were lysed in the presence of 500 M phosphopeptide 612 or 750 or 100 M 1009 and precipitations prepared with the anti-SHP2 antibody. Phosphopeptides 612 and 1009 significantly reduced the co-precipitation of endogenous SHP2 with tyrosine-phosphorylated ␤c from IL-3-stimulated cells (Fig. 3D, lanes 3 and 5), whereas phosphopeptide 750 did not. It is interesting that although SHP2 is inhibited from binding to ␤c by phosphopeptides 612 and 1009, the level of SHP2 tyrosine phosphorylation is not decreased and in this experiment appears to have increased (72-kDa protein, Fig. 3D,  compare lanes 3 and 5 with lanes 2 and 4). Fig. 3D, lower panel, shows the blot to be evenly loaded with respect to SHP2. This result suggests that association of SHP2 with ␤c may not be required for its tyrosine phosphorylation. Higher concentrations of phosphopeptide were required to inhibit co-precipitation of ␤c by anti-SHP2 antibodies compared with the fusion proteins, perhaps reflecting a high affinity complex between SHP2 and ␤c. Competition experiments were attempted using the anti-SHP1 antibody, but proved technically challenging. The anti-SHP1 antibody appears to be much less efficient in immunoprecipitation (even though 10 times more was used than for the SHP2 antibody), and the amount of the tyrosinephosphorylated ␤c precipitated was often low (see Fig. 1A, lane  6), making interpretation of the results difficult. This could be explained if the association between SHP1 and the receptor is Phosphopeptide 612 Is a Substrate for SHP1 and SHP2-To assess whether the synthetic phosphopeptides tested in the competition experiments could also serve as substrates for SHP1 and SHP2, phosphatase assays were performed. GSTfusion proteins of SHP1 and SHP2 lacking the SH2 domains were cleaved with thrombin to release the GST-protein part. However, due to a cryptic thrombin cleavage site within the carboxyl terminus of SHP1 a 5-kDa fragment was also released after complete thrombin digestion (42). The activity of both PTPases, SHP1(⌬SH2/⌬CT), and SHP2 (⌬SH2) toward the five different phosphopeptides was assessed, and the results are shown in Fig. 4. With both enzymes, phosphopeptide 612 was by far the best substrate as reflected in the highest release of P i . Phosphopeptide 750 showed a much diminished, but significant, release of P i , whereas with phosphopeptides 695, 577, and 452, either no or minimal dephosphorylation was seen. Thus, a phosphopeptide corresponding to a potential binding site for the SH2 domains of both enzymes serves as a good substrate for the PTPase domain of both SHP1 and SHP2. DISCUSSION In this study we have demonstrated that the tyrosine phosphatases SHP1 and SHP2 can both bind inducibly to ␤c following IL-3 stimulation. This association appears to be directly mediated by interactions between the SH2 domains of SHP1 and SHP2 and phosphotyrosine residues within ␤c. A phosphotyrosine-containing peptide based on the sequence surrounding tyrosine 612 of ␤c was able to compete the binding of both SHP1SH2-GST and SHP2SH2-GST fusion proteins to tyrosinephosphorylated ␤c in in vitro assays and also the binding of endogenous SHP2 to ␤c in immunoprecipitation studies. These results strongly suggest that the SH2 domains of both SHP1 and SHP2 bind to tyrosine 612 of ␤c.
Tyrosine 612 of ␤c is located within the motif LEYLCLP, which has similarities to motifs previously identified for SHP1 and SHP2 SH2 interactions. The NH 2 -terminal SH2 domain of SHP1 showed a broad selectivity for pY-hydrophobic-X-hydrophobic motifs from a phosphopeptide library (49). In addition, tyrosine 429 of the EpoR, which lies in the pYLYL motif, has been show to be essential for the SHP1 binding (18). Pei et al. (50) have previously shown that a phosphopeptide based on the sequence surrounding Tyr 612 of ␤c (referred to as Tyr 628 , which includes the 14-amino acid signal peptide) bound the NH 2terminal SH2 domain of SHP1, activated the phosphatase, and acted as a substrate for SHP1. They suggested this tyrosine may be the binding site for SHP1 to ␤c (50), and our results show a similar peptide does compete with SHP1 for binding to ␤c. We observed weaker inhibition of precipitation of ␤c with SHP1SH2-GST by phosphopeptide 750. Tyrosine 750 is located in the sequence pYVEL, which also conforms to the predicted SHP1 SH2 binding motif (49).
The selectivity of the NH 2 -terminal SH2 domain of SHP2, determined using a degenerate peptide library, was shown to be pY-V/I/T-X-V/L/I (51). Experiments using both mutant receptors (31) and peptide competition assays (27) demonstrate that tyrosine 1009 of the PDGFR, in the motif pYTAV, is required for SHP2 binding. Recently, using EpoR mutant receptors, it has been shown that SHP2 binds, via its SH2 domains, to the activated EpoR at tyrosine 425, in the motif pYTIL (32). The residues surrounding tyrosine 612 (YLCL) of ␤c are similar to these previously described binding motifs for SHP2.
The effects of mutagenesis of tyrosines 612 and 750 of ␤c on tyrosine phosphorylation of substrates in response to GM-CSF have been reported (52,53). In these transfectants, normal levels of SHP2 tyrosine phosphorylation were observed (53). However, the association of SHP2 with ␤c was not examined in these mutant ␤c-expressing cells. An interesting point relating to this is whether stable association of SHP2 with ␤c is required for its tyrosine phosphorylation. Our data suggest that in the presence of phosphopeptide 612, which competes for the binding of SHP2 to ␤c, the levels of SHP2 tyrosine phosphorylation are not diminished and may actually increase (Fig. 3D), supporting the notion that SHP2 does not need to be bound to ␤c to become phosphorylated. However, we cannot rule out the possibility that SHP2 associates transiently with ␤c and that this is all that is required for its phosphorylation or that another mechanism is leading to increased SHP2 tyrosine phosphorylation in the presence of the phosphopeptide, e.g. a kinase is activated. Additional mutational analyses of ␤c have implicated two regions of ␤c, which appear to influence SHP2 tyrosine phosphorylation (54). Mutation of Tyr 577 in conjunction with a truncation up to residue 589 resulted in greatly reduced levels of SHP2 tyrosine phosphorylation in response to GM-CSF, although either mutation alone had no effect (54). Our results suggest that Tyr 577 is not involved in SHP2 binding to ␤c and that Tyr 612 is the major site of interaction. Since Tyr 612 is removed in the 589 ␤c truncation mutant, but SHP2 is still tyrosine-phosphorylated, it may be that association of SHP2 is not required for its tyrosine phosphorylation, and our data are Interestingly, phosphotyrosine peptide 612, which competed the binding of both SHP1 and SHP2 to ␤c, was also the best substrate for SHP1 and SHP2 catalytic activities. It has been shown previously that SHP1 and SHP2 prefer substrates that have acidic residues to the NH 2 terminus of the phosphotyrosine (42,55), and this is the case for Tyr 612 . Whether this site is an in vivo substrate remains to be determined, although Yi et al. (17) reported the SHP1-catalyzed dephosphorylation Aic2A. Therefore, the same site that appears to be recognized by the SH2 domains of these phosphatases can also be utilized as a substrate. In experiments using full-length recombinant SHP2 and peptides for substrates, Sugimoto et al. (55) found SHP2 to have a preference for tyrosine 1009 (to which the SH2 domain of SHP2 also binds, leading to its activation (30)) and tyrosine 1021 of the PDGFR, suggesting that tyrosine 1009 may both regulate and act as a substrate for the PTPase activity of SHP2. We have observed similar substrate specificities with PDGF receptor peptides (42), and the results presented here with SHP2 and ␤c suggest a similar mechanism may be used in IL-3 signaling. However, using immunoprecipitated phosphorylated PDGF␤ receptor and recombinant full-length SHP2, phosphotyrosines Tyr 771 and Tyr 751 , followed by Tyr 740 , were used preferentially, while Tyr 1021 and Tyr 1009 were very poor substrates (56). These discrepancies could arise from differences in using the intact PDGF ␤ receptor, which contains multiple potential phosphorylation sites, which may bind other proteins and mask potential sites, instead of using peptides as substrates. In our in vitro assay system, the SHP1 and SHP2 recombinant enzymes used lacked their SH2 domains, thus removing any potential "activating" effects of the various phosphopeptides.
It is likely that SHP1 and SHP2 have different roles when localized to the IL-3 receptor. SHP1 is thought to be a negative regulator of growth and functions to terminate signals. Overexpression of SHP1 in DA3 cells leads to a decrease in IL-3-dependent proliferation and a decrease in Aic2A tyrosine phosphorylation (17). The binding of SHP1 to the EpoR activates the PTPase, which dephosphorylates Jak-2, leading to the termination of proliferative signals (18). SHP1 has been shown recently to interact directly with Jak-2, leading to its dephosphorylation (57). IL-3 also induces activation of Jak-2 (5), so SHP1 may function in a similar manner in IL-3 signal transduction. SHP1 may also function to dephosphorylate other phosphorylated tyrosine residues on the IL-3 receptor, again leading to down-regulation of IL-3 responses. However, also competing for the same binding site on ␤c is SHP2, which is thought to act as a positive mediator of growth factor signals. SHP2 associates with Jak-1 and Jak-2, and tyrosine 304 of SHP2 is phosphorylated by these kinases, leading to the creation of a Grb2 recognition motif (58). We have shown previously that tyrosine-phosphorylated SHP2 associates with the adaptor molecule Grb2 (7), and the studies presented here suggest that SHP2 could act as an adaptor between the activated ␤c and Grb2, thus leading to activation of the ras/mitogen-activated protein kinase pathway, known to be activated by IL-3 (9,59). This has been suggested in other systems, where SHP2 acts as an adaptor between Grb2 and the EpoR (28) and the PDGFR (38). Expression of a catalytically inactive version of SHP2 results in reduced activation of the mitogenactivated protein kinases Erk-1 and Erk-2 in response to insulin (60), which fits with the proposed role of SHP2 in transducing positive growth-promoting signals. Additionally, we have shown previously that SHP2 associates with the p85 subunit of phosphatidylinositol 3Ј-kinase (7), so, SHP2 may also regulate this pathway. Although both SHP1 and SHP2 associate with tyrosine 612 of ␤c, it is likely that only a portion of the receptors are associated with SHP1 and SHP2 at any one time. The kinetics of activation/deactivation of each phosphatase would have to be individually assessed to determine the relative sequence of events. Any shifting of the equilibrium between the two phosphatases would result in either a positive or negative effect on IL-3-induced signals. Further detailed molecular analyses are required to dissect these pathways.
In summary, we have demonstrated that SHP1 and SHP2 bind through their SH2 domains to tyrosine 612 of ␤c. Both SHP1 and SHP2 appear to be able to regulate their binding to the receptor as phosphopeptide 612 also served as a substrate for the catalytic domain of both the PTPases, possibly enabling them to modulate signaling pathways, regulated by tyrosine phosphorylation/dephosphorylation events.