Shc interaction with Src homology 2 domain containing inositol phosphatase (SHIP) in vivo requires the Shc-phosphotyrosine binding domain and two specific phosphotyrosines on SHIP.

The adapter protein Shc has been implicated in mitogenic signaling via growth factor receptors, cytokine receptors, and antigen receptors on lymphocytes. Besides the well characterized interaction of Shc with molecules involved in Ras activation, Shc also associates with a 145-kDa tyrosine-phosphorylated protein upon triggering via antigen receptors and many cytokine receptors. This 145-kDa protein has been recently identified as an SH2 domain containing 5'-inositol phosphatase (SHIP) and has been implicated in the regulation of growth and differentiation in hematopoietic cells. In this report, we have addressed the molecular details of the interaction between Shc and SHIP in vivo. During T cell receptor signaling, tyrosine phosphorylation of SHIP and its association with Shc occurred only upon activation. We demonstrate that the phosphotyrosine binding domain of Shc is necessary and sufficient for its association with tyrosine-phosphorylated SHIP. Through site-directed mutagenesis, we have identified two tyrosines on SHIP, Tyr-917, and Tyr-1020, as the principal contact sites for the Shc-phosphotyrosine binding domain. Our data also suggest a role for the tyrosine kinase Lck in phosphorylation of SHIP. We also show that the SH2 domain of SHIP is dispensable for the Shc-SHIP interaction in vivo. These data have implications for the localization of the Shc.SHIP complex and regulation of SHIP function during T cell receptor signaling.

Shc interaction with activated receptors can occur either via the SH2 or the PTB domain. While both domains bind phosphotyrosine-containing sequences, their specificities of recognition are different and require residues either COOH-terminal (for SH2) or NH 2 -terminal (for PTB) to the Tyr(P) (28 -33). In hematopoietic cells, the Shc SH2 domain binds to components of the T cell receptor (TCR) (6,11) and B cell receptor (34), while the PTB domain interacts with the receptors for interleukin (IL)-2 (35) and IL-3 (36). Structures of the Shc PTB and Shc SH2 domains bound to their respective phosphopeptides have revealed the molecular basis for their interactions.
Besides interaction with the activated receptors, Shc also associates with a prominent 145-kDa tyrosine-phosphorylated protein upon activation of several receptors on hematopoietic cells (such as the TCR, B cell receptor, receptors for IL-2, IL-3, granulocyte-macrophage colony-stimulating factor (CSF), erythropoietin, steel factor, and macrophage CSF) (12,13,17,37,38). Recently, this 145-kDa phosphoprotein has been identified as a novel SH2 domain containing 5Ј-inositol phosphatase, SHIP (39 -41). SHIP has homology to several previously identified 5Ј-inositol phosphatases that have been linked to certain congenital metabolic disorders (42,43). Moreover, SHIP has been shown to dephosphorylate the 5Ј position of phosphatidylinositol 3,4,5-P 3 and inositol 1,3,4,5-tetrakisphosphate (39 -41). The precise role of SHIP in regulating inositol lipids in vivo and the influence of these phosphoinositides (generated by SHIP-mediated dephosphorylation) on downstream signaling events remain unclear. SHIP has been implicated in inhibitory signaling via Fc␥RIIb on mast cells (44) and mitogenic signaling via macrophage CSF (41). Since 3Ј-OH-phosphorylated inositol lipids are generated by the triggering of many receptors (through the action of the enzyme phosphatidylinositol 3-kinase) (45), the dephosphorylation of phosphatidylinositol 3,4,5-P 3 by SHIP could down-regulate signals via phosphatidylinositol 3,4,5-P 3 ; alternatively, phosphatidylinositol 3,4-P 2 , the product of the SHIP-mediated dephosphorylation, could lead to activation of kinases such as Akt (45). However, the mechanisms by which SHIP regulates signaling via these receptors are not yet understood.
In vivo, SHIP may be regulated through alteration of its catalytic activity or through its association with other proteins such as Shc. The initial studies reported that, during macrophage CSF and IL-3 signaling (39,41), there was no detectable alteration of the catalytic activity of SHIP following receptor activation. This suggested that the interaction of SHIP with Shc and, in turn, the subcellular localization of the Shc⅐SHIP complex may be quite important in SHIP-mediated regulation of specific phosphoinositides. The details of the in vivo molecular interaction between Shc and SHIP have not been elucidated. Some in vitro studies indicated a direct binding of the PTB domain to phosphorylated SHIP (21,37). However, Liu et al. (46), based on in vitro studies, have proposed that the SH2 domain of SHIP may also reciprocally bind to phosphorylated Shc. In this report, we have addressed the in vivo requirements for the interaction between Shc and SHIP during T cell receptor signaling. Our data show that the Shc-PTB domain and specific tyrosine residues within the COOH terminus of SHIP mediate their association in vivo.
Plasmids-DNA encoding GST-tagged full-length (FL) wt Shc, FL Shc R175Q, PTB domain alone (Shc-PTB), or the CH and SH2 domains of Shc (Shc CH-SH2) were generated by polymerase chain reaction and subcloning of DNA fragments into the pEBG vector, as described previously (36,48). GST-tagged SHIP constructs were also generated using the same strategy, and a polymerase chain reaction-based mutagenesis was used to generate the various mutants (36,49). To generate hemagglutinin (HA)-tagged constructs, oligonucleotides encoding three HA tags in tandem were cloned into the pEBB vector (48), and the DNA encoding different SHIP regions were subcloned in-frame at the 3Ј end of the tag. All constructs were sequenced, and the presence of appropriate mutations was confirmed. The Lck cDNA cloned into pEF vector was kindly provided by Dr. Tomas Mustelin.
Antibodies-Antibody specific for the SH2 domain of SHIP has been described previously (39,44). Antibodies specific for Shc and GST were purchased from Transduction Laboratories (Kentucky) and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. Horseradish peroxidaselabeled anti-Tyr(P) antibody (RC20H) was purchased from Transduction Laboratories. Anti-HA antibody 12CA5 was obtained from Boehringer Mannheim.
Transfections-Transient transfections into COS cells were performed with DEAE-dextran and chloroquine as described previously (36) with 1 g each of the indicated DNA. Approximately 24 h posttransfection, the cells were starved overnight without serum and harvested, and the proteins were analyzed as described below. Transient transfections into BYDP cells (1 ϫ 10 7 /ml in growth medium) were performed by electroporation using 50 g of the indicated DNA (Cell Porator, Life Technologies, Inc.) at 250 V and 1180 microfarads. The transfected cells (after approximately 24 h) were stimulated and processed as described below. BYDP cells stably expressing Shc proteins were generated by using the electroporation conditions described above with 20 g of the relevant plasmids co-transfected with 5 g of pMHneo plasmid for selection (50). Forty-eight hours post-transfection, the cells were plated out at 2 ϫ 10 4 cells/well in 48-well plates in medium containing G418 (Life Technologies, Inc.). Individual clones were screened by Western blotting for expression of the proteins of interest, and positive clones that expressed roughly equivalent levels of the different proteins were used in further studies.

RESULTS AND DISCUSSION
Identification of p145 as SHIP-We have previously observed (11, 35, 37) that a 145-kDa tyrosine-phosphorylated FIG. 1. Identification of Shc-associated p145 as SHIP. a, murine T cell hybridoma cells were activated by cross-linking the TCR with CD4 (TCR ϫ CD4) as described under "Experimental Procedures." SHIP and Shc proteins were immunoprecipitated and analyzed by anti-Tyr(P) (anti-pTyr) or anti-SHIP immunoblotting. Ipt, immunoprecipitate. b, T cells were stimulated by cross-linking CD4 alone, TCR alone, or TCR ϫ CD4 for 2 min at 37°C. The lysates were precipitated with anti-SHIP and immunoblotted with anti-Tyr(P) antibody (RC20H). The levels of SHIP precipitated in different lanes was determined by the stripping and reprobing of the same blot with the anti-SHIP antibody.
protein co-precipitates with Shc upon TCR and IL-2 receptor stimulation in T cells. We first determined if this p145 protein represents SHIP (17,40,41,44). Immunoprecipitation of Shc following cross-linking of TCR with CD4 and immunoblotting with antibodies specific for SHIP demonstrated that the 145-kDa phosphoprotein is in fact SHIP (Fig. 1a). This band comigrated with SHIP proteins detected in direct anti-SHIP immunoprecipitates (Fig. 1a). Two additional observations were made. First, SHIP associated with Shc only upon activation. Second, very little tyrosine phosphorylation of SHIP was detected in unstimulated T cell lysates. This is in contrast to the previous observations in IL-3-dependent cells where there was considerable basal phosphorylation of SHIP, which was augmented by IL-3 stimulation (39). We observe a similar basal phosphorylation of SHIP in the IL-2-dependent cell line CTLL-20 (data not shown). Thus, different cell types may have variations in the levels of basal tyrosine phosphorylation of SHIP, which may have implications for SHIP functional activity. However, it is important to note that despite the basal tyrosine phosphorylation, SHIP associated with Shc only upon IL-2 and IL-3 stimulation (data not shown and Ref. 39).
We also observed that cross-linking of the TCR with the co-receptor CD4 enhances the tyrosine phosphorylation of SHIP. As shown in Fig. 1b, cross-linking of CD4 alone or cross-linking of TCR alone led to detectable tyrosine phosphorylation of SHIP, while activation via both receptors significantly enhanced the phosphorylation on SHIP. Since CD4 is physically associated with the tyrosine kinase Lck and the Lck kinase activity is enhanced by CD4 cross-linking, Lck may play a role in phosphorylation of SHIP, either directly or indirectly.
Consistent with this notion, co-expression of Lck with SHIP leads to efficient tyrosine phosphorylation of SHIP in COS cells (see below).
Shc-PTB Domain Is Required for Interaction with SHIP in Vivo-Shc contains two domains capable of interacting with tyrosine-phosphorylated proteins: an amino-terminal PTB domain and a COOH-terminal SH2 domain (7,20,21). To determine which region(s) of Shc is necessary for interaction with SHIP, we generated stable murine T cell lines expressing either the Shc-PTB domain alone (GST-Shc-PTB) or the CH and SH2 domains of Shc (GST-Shc CH-SH2), tagged at the amino terminus with GST (Fig. 2a). Cells expressing comparable amounts of the GST-tagged Shc proteins or control cells expressing the GST tag alone were used in all of the studies described below, and the data shown are representative of several independent clones derived from multiple transfections. When cells expressing the GST-Shc-PTB were activated via the T cell receptor and the cell lysates were precipitated with glutathione-Sepharose beads, SHIP was co-precipitated with Shc-PTB only upon activation (Fig. 2b). In control cells, the GST tag alone did not precipitate SHIP. This is consistent with our previous in vitro studies (35) using bacterially expressed Shc proteins which had indicated a PTB-dependent interaction with SHIP. Comparison of cells expressing the PTB or the CH-SH2 domains of Shc indicated that SHIP co-precipitates only with the PTB domain and not with the CH or SH2 domains of Shc (Fig. 2c). Thus, we conclude that in vivo, the Shc-PTB domain (and not the SH2 domain) is essential for the interaction with SHIP during T cell receptor signaling. . c, cells expressing GST alone, Shc-PTB, or Shc CH-SH2 domains were stimulated as above, and the lysates were precipitated with GSH beads. The bound proteins were immunoblotted with antibodies to SHIP. Anti-SHIP immunoprecipitation is shown for comparison. The presence of all GST-tagged proteins was confirmed by anti-GST-immunoblotting (data not shown). Ipt, immunoprecipitates.

SH2 Domain of SHIP Is Dispensable for Interaction with
Shc-In the Shc⅐SHIP complexes that occur upon receptor activation, both Shc and SHIP are tyrosine-phosphorylated. With the identification of an SH2 domain in SHIP (39 -41), a second interaction involving SHIP-SH2 binding to phosphorylated Shc has also been proposed (39). In fact, Damen et al. (39) have reported that a phosphopeptide containing the Tyr-317 site of Shc (attached to beads) can precipitate SHIP from cell lysates. While our above data using T cell lines expressing individual domains of Shc clearly demonstrated the requirement for the Shc-PTB domain, it did not exclude a parallel role for the SHIP-SH2 domain in binding to Shc. We examined if this interaction occurs in vivo and determined its relative significance compared with the Shc-PTB binding to phosphorylated SHIP.
To specifically test the role of SHIP-SH2, we needed a setting where the Shc proteins will be phosphorylated, yet the PTB domain would no longer be able to interact with SHIP. The rationale was that if SHIP-SH2 does play a role in binding to phosphorylated Shc, then SHIP would co-precipitate with a phosphorylated PTB-defective Shc mutant. We and others (36,49,51) have previously shown that mutation of Arg-175 within the PTB domain abrogates binding to Tyr(P)-containing proteins. Using a transient transfection approach in COS cells, we expressed HA-tagged wt SHIP proteins along with GST-tagged FL wt Shc or FL Shc R175Q. To obtain phosphorylation of the expressed Shc and SHIP proteins, we coexpressed the T cell tyrosine kinase Lck under these conditions. GST-Shc was precipitated, and the co-precipitation of HA-SHIP was examined by both anti-Tyr(P) and anti-HA immunoblotting.
As shown in Fig. 3, the HA-SHIP proteins were co-precipitated efficiently by wt Shc but not by the R175Q Shc. This again highlighted the role for the PTB domain in this interaction. Moreover, as shown in Fig. 3 (top panel), the Shc wt and the Shc R175Q mutant were comparably tyrosine-phosphorylated. This finding is consistent with our previous data that Tyr-317 appears to be the major, if not the only, site of tyrosine phosphorylation on Shc during T cell activation and that this site is efficiently phosphorylated by Lck in vitro (37) and in vivo. 2 However, despite the phosphorylation on the FL Shc R175Q, SHIP failed to interact with this protein, thereby suggesting lack of a role for the SHIP-SH2 domain in binding to phosphorylated Shc. Furthermore, when the Shc-PTB or CH-SH2 domains were coexpressed with HA-SHIP and Lck, the CH-SH2 domain (which contains Tyr-317), despite its tyrosine phosphorylation, failed to co-precipitate SHIP. On the other hand, the PTB domain, despite the lack of phosphorylation, efficiently co-precipitated SHIP. Comparable expression of Shc and SHIP proteins was confirmed by anti-GST or anti-HA immunoblotting (Fig. 3, 2 bottom panels). Based on these data we conclude that the SHIP-SH2 does not contribute to the Shc-SHIP interaction in vivo. We cannot exclude very weak or transient interactions (that do not withstand the precipitation conditions) that may occur between SHIP-SH2 and phosphorylated Shc. However, this clearly cannot be responsible for the strong interaction that is observed between Shc and SHIP. Thus, our data demonstrate that the interaction of Shc-PTB with tyrosine-phosphorylated SHIP is the primary mode of interaction between these two proteins in vivo.
Two Tyrosines (Tyr-917 and Tyr-1020) on SHIP Mediate Shc-PTB Binding-We then determined which tyrosine(s) on SHIP provided the binding site(s) for the Shc-PTB. The preferred binding motif for the Shc-PTB domain has been defined as ⌿XNXXY (where ⌿ represents hydrophobic residues) through analysis of phosphopeptide binding to Shc-PTB (28, 30 -33). Based on this motif, two tyrosines in the COOH-terminal region of SHIP, Tyr-917 and Tyr-1020, were identified as putative Shc-PTB binding sites. To determine if these two tyrosines on SHIP contribute to Shc binding, individual or combined mutations of Tyr-917 and Tyr-1020 to phenylalanine were performed (Fig. 4a). wt SHIP, Y917F, Y1020F, or Y917F/ Y1020F double mutant SHIP proteins (all tagged with GST) were transiently expressed in the murine hybridoma cells. After T cell activation, the expressed proteins were precipitated using glutathione-Sepharose beads, and the tyrosine phosphorylation status of these proteins was assessed by anti-Tyr(P) immunoblotting. As shown in Fig. 4b, compared with wt SHIP, the phosphorylation of Y917F and Y1020F proteins was decreased moderately, while phosphorylation of the Y917F/ Y1020F double mutant was not detectable. These data suggested that these two tyrosines may represent key phosphorylation sites on SHIP during T cell receptor activation. However, we were unable to monitor co-precipitation of Shc in these experiments due to the insufficient expression 2 K. S. Ravichandran, unpublished observations. To test the role of these two tyrosines in greater detail, we transiently expressed HA-tagged wt and mutant SHIP proteins along with GST-tagged Shc-PTB in COS cells. The tyrosine kinase Lck was again coexpressed to mediate phosphorylation of SHIP. The cells were lysed, and the Shc-PTB was precipitated using glutathione-Sepharose beads. The co-precipitating SHIP proteins were visualized by anti-HA immunoblotting. As shown in Fig. 4c, wt SHIP bound to Shc quite efficiently. However, SHIP proteins with a mutation of either Tyr-917 or Tyr-1020 showed a significant decrease in binding to Shc-PTB, while the Y917F/Y1020F double mutation nearly abrogated binding to Shc-PTB. In contrast, deletion of either the SH2 domain of SHIP (⌬SH2) or a point mutation within the SH2 domain of SHIP (R34Q, within the conserved FLVR sequence) had little effect on their interaction with Shc-PTB. Anti-HA immunoblotting of total lysates indicated that all of the mutant HA-SHIP proteins were expressed (Fig. 4c, bottom panel). Direct anti-Tyr(P) blotting of HA-SHIP precipitates indicated a diminished, yet detectable, phosphorylation of the Y917F/ Y1020F double mutant. Although tyrosine phosphorylation of the Y917F/Y1020F double mutant was not detectable in transient expression in T cells (Fig. 4b), phosphorylation of this protein was observed in COS cells (Fig. 4d). This may be due to the apparently high activity of the Lck when expressed in COS cells, leading to phosphorylation on tyrosines other than Tyr-917 and Tyr-1020 on SHIP. Also, the apparently increased phosphorylation of the Y917F mutant (Fig. 4d) can be ex- FIG. 4. Tyr-917 and Tyr-1020 mediate Shc-PTB binding to SHIP in vivo. a, schematic diagram of different SHIP constructs. IPase, inositol phosphatase. b, the murine T cell hybridoma cells (BYDP) were transiently transfected with the indicated GST-tagged SHIP constructs. 24 h post-transfection the cells were stimulated as above and lysed, and the proteins were precipitated with GSH beads. The phosphorylation on different SHIP proteins was analyzed by anti-Tyr(P) (anti-pTyr) immunoblotting using the monoclonal antibody 4G10. The blot was stripped and reprobed with anti-GST to reveal the level of SHIP protein in different lanes (bottom panel). Ppt, precipitate. c, COS cells were transiently transfected with GST-tagged Shc-PTB and Lck along with the indicated HA-tagged SHIP constructs. The lysates were precipitated with GSH beads, and the level of co-precipitating SHIP was determined by anti-HA immunoblotting. The levels of precipitated Shc-PTB in different lanes (middle panel) as well as the expression of the different HA-tagged SHIP proteins in total lysates (bottom panel) were determined by anti-GST and anti-HA blotting, respectively. d, tyrosine phosphorylation of the different HA-tagged SHIP proteins expressed in COS cells (simultaneously expressed with Lck and Shc-PTB, as in c above) was determined by direct anti-HA precipitation followed by anti-Tyr(P) immunoblotting (top panel). The bottom panel indicates the level of HA-tagged SHIP protein in different lanes. ppt, precipitate; Ipt, immunoprecipitate. plained by the higher expression of this protein in this experiment. All of the mutant proteins had 5Ј-inositol phosphatase activity, indicating the mutations did not induce gross structural changes of the protein (data not shown). Thus, we conclude that Tyr-917 and Tyr-1020 represent the two critical tyrosines within SHIP whose phosphorylation is required for the efficient interaction with Shc in vivo.
The data presented in this report have several implications for signaling via Shc and SHIP. In the case of Fc␥RIIB, SHIP has been shown directly to bind to the receptor via its SH2 domain (44). However, we do not observe a direct binding of SHIP to either the TCR complex or the interleukin-2 receptor (data not shown). Under these conditions, the interaction of SHIP with proteins such as Shc may play a much greater role in its localization. At a biochemical level, the interaction between Shc and SHIP could occur via more than one mechanism, i.e. Shc-PTB binding to phosphorylated SHIP and/or the SHIP-SH2 binding to phosphorylated Shc. The data presented in this report demonstrate that in T cells and in COS cells, the Shc-PTB domain binding to phosphorylated SHIP (specifically Tyr-917 and Tyr-1020) is necessary and sufficient for this interaction. Our data also suggest that both sites (Tyr-917 and Tyr-1020) can contribute to Shc binding. Although Liu et al. (46), based on peptide competition studies, have indicated a SHIP-SH2 interaction with phosphorylated Shc, we do not observe this interaction in vivo under the conditions we have tested. This suggested that the SH2 domains of both proteins are "free" to interact with other molecules.
In the case of the TCR, Shc binds directly via its SH2 domain to the TCRchain (11); in this case, Shc, via its PTB domain, may recruit SHIP to the proximity of the TCR. In our preliminary studies we have failed to observe a co-precipitation of SHIP with components of the TCR⅐CD3 complex, and the existence of such a trimeric complex during T cell activation remains to be determined. Since the SHIP-SH2 domain does not participate in binding to Shc, an interesting possibility is that the SHIP-SH2 may be free to bind other proteins, which may influence its function. In the case of the IL-2 receptor, Shc directly binds to the IL-2 receptor ␤ chain via its PTB domain (35). Structural studies of the PTB domain indicate the existence of only a single Tyr(P) binding pocket within the Shc-PTB (49). Thus, the Shc⅐IL-2 receptor and Shc⅐SHIP complexes must exist separately and may perform independent roles. Since Shc interacts via its PTB domain with the ␤ c chain (shared by IL-3, IL-5, and granulocyte-macrophage CSF receptors) (36), similar independent complexes may exist during signaling via these cytokines as well. Thus, depending on the type of receptor, recruitment of SHIP to active signaling complexes may occur by different mechanisms and, in turn, may have different functional consequences.