Grb10 Interacts Differentially with the Insulin Receptor, Insulin-like Growth Factor I Receptor, and Epidermal Growth Factor Receptor via the Grb10 Src Homology 2 (SH2) Domain and a Second Novel Domain Located between the Pleckstrin Homology and SH2 Domains*

The Grb10 protein appears to be an adapter protein of unknown function that has been implicated in insulin receptor (IR) signaling. The interaction of this protein with the IR has been shown to be mediated in part by the Src homology 2 (SH2) domain of Grb10. Here we demonstrate the existence of a second novel domain within Grb10 that interacts with the IR and insulin-like growth factor receptor in a kinase-dependent manner. This domain was localized to a region of approximately 50 amino acids, and we term it the BPS domain to denote its location between the PH and SH2 domains. The BPS domain does not bear any obvious resemblance to other known protein interaction domains but is highly conserved among the Grb10-related proteins Grb7 and Grb14. We show that the BPS domain interaction is de-pendent upon receptor tyrosine kinase activity. Fur-thermore, interaction of the BPS domain requires the kinase domain of the IR, since mutation of the paired tyrosine residues (Y1150F/Y1151F) within the IR activation loop dramatically reduced the interaction. Last, our data suggest that the presence of two distinct protein interaction domains may help to determine the specificity by which Grb10 interacts with different receptors. Specifically, the IR, which appears to interact most strongly with Grb10, interacts well with both the SH2 and BPS domains. Conversely, the insulin-like solution of 3-bromo-5 9 -deoxyuridine (BrdUrd) (Amersham Pharmacia Biotech) medium 16 h to label newly synthe- sized period, fixed and processed for immunofluorescence as described by Xiao et al. (9). Briefly, stained with a rat monoclonal anti-BrdUrd antibody (Accurate Scientific), followed by a tetramethylrhodamine B isothiocyanate-conjugated donkey anti-rabbit IgG antibody and a fluo-rescein isothiocyanate-conjugated donkey anti-sheep IgG antibody to detect the injected cells. This technique results in nuclear staining in have S and analyzed and quantitated

The binding of hormones to their receptor tyrosine kinases triggers receptor tyrosine autophosphorylation (1). In the case of the insulin receptor (IR) 1 and insulin-like growth factor I receptor (IGFIR), phosphorylation of multiple tyrosine residues leads to activation of receptor kinase activity (2) and, in addition, creates docking sites for downstream adaptor proteins such as SHC, insulin receptor substrate-1 (IRS-1), and insulin receptor substrate-2 (IRS-2) (3)(4)(5). Phosphorylation of IRS-1 upon multiple tyrosines results in its association with Src homology 2 (SH2) domain-containing proteins including Grb2, Syp, NCK, and the p85 subunit of phosphatidylinositol 3-kinase, thus activating various signaling cascades (6 -10). Tyrosyl phosphorylation of SHC also leads to its interaction with Grb2 and mediates the activation of the guanine nucleotide exchange factor SOS and subsequent activation of the Ras signaling pathway (11)(12)(13)(14).
SHC, IRS-1, and IRS-2 proteins have been shown to interact with the IR or the related IGFIR through phosphotyrosine binding domains, which recognize the phosphorylated NPXY motif in the juxtamembrane domains of receptors (3,5,15). In addition, a second poorly understood receptor binding domain has also been identified within the central domain of IRS-2 (5,16). Both SHC and IRS-1 have been implicated in insulinstimulated mitogenic signal transduction (14,(17)(18)(19), and IRS-1 has also been proposed to mediate some of the metabolic effects of the IR, including regulation of glucose transport partly through the activation of phosphatidylinositol 3-kinase (20 -22). Whether these proteins are sufficient to explain the diverse effects of insulin receptor signaling is a matter of intense investigation. In this regard, the recent discovery of the related proteins that interact with the IR termed Grb-IR, Grb10/IR-SV1, or GrbIR␤/Grb10 by several groups is of interest (23)(24)(25). These three Grb10 variants are derived from the same gene via alternative splicing and have recently been renamed Grb10␣ for Grb-IR and Grb10␤ for Grb10/IR-SV1 and Grb-IR␤/Grb10, the latter two being identical cDNAs. Grb10␣ contains a distinct amino terminus compared with Grb10␤, and in addition, Grb10␤ contains an intact PH domain, which is partially deleted due to a splicing event in the Grb10␣ mRNA. Since the Grb10 domains that we are addressing in this paper are found in both the ␣ and ␤ forms of Grb10, for simplicity we will refer to the protein as Grb10 in the remainder of the paper.
Grb10 (26) is a member of a family of proteins that also * This work was supported by National Institutes of Health Grants DK44093 and DK50602 (to T. A. G.) and Special Research Initiative Support from the University of Maryland. This work was also supported in part by National Institutes of Health Grant DK33651 and by the Veterans Administration Medical Research Service (to J. M. O.) 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.
ʈ Supported by an American Diabetes Association Career Development Award.
includes Grb7 (27,28) and Grb14 (29). These proteins share highly homologous amino acid sequences and common domains, including a C-terminal SH2 domain, a conserved proline-rich region within the N terminus, and a ϳ300-amino acid conserved central domain that includes the PH domain and other conserved regions of unknown structure or function located both upstream and downstream of the PH domain (26,28,29). The Grb7/Grb10/Grb14 proteins were originally cloned by screening cDNA expression libraries with a labeled tyrosylphosphorylated carboxyl-terminal fragment of the EGFR (26,28,29). However, neither Grb10 nor Grb14 have been demonstrated to interact efficiently with the EGFR in cells (26,29). Grb10 has also been identified using yeast two-hybrid methodologies using the cytoplasmic domains of the IR or IGFIR as "bait" (23,24,30,31), and it has recently been proposed that Grb10 associates preferentially with the IR as compared with the IGFIR in mouse fibroblasts (30). The role that Grb10 plays in cellular signaling mediated by any of these receptor tyrosine kinases remains unclear, as do the roles of Grb7 and Grb14. The multidomain structure of these proteins suggests that they act as adapter molecules, but the signaling pathways that are affected and the nature of this effect remain to be identified.
It has previously been shown that the SH2 domain located at the extreme C terminus of Grb10 is important for interaction with the IR (23,25,31,32). These data are consistent with the previous demonstrations that (a) Grb10 interaction depends upon IR kinase activity in the two-hybrid assay, (b) in vitro interaction of the IR with Grb10-GST fusion proteins is insulindependent, and (c) coimmunoprecipitation of Grb10 with the IR is insulin-dependent (23-25, 30 -32). However, some recent evidence has suggested that the SH2 domain of Grb10 may not fully explain the interaction of Grb10 with the IR. Specifically, Frantz et al. showed that a full-length Grb10-GST fusion protein was able to interact more efficiently with the IR in vitro than the SH2 domain alone (25). These authors suggested that the PH domain might play a role in mediating the Grb10 interaction, since a construct containing both the PH and SH2 domains showed more efficient interaction. To address this issue, we used the yeast two-hybrid system and assays of in vitro protein interaction to demonstrate that Grb10 does indeed contain a second domain capable of independently interacting with the IR in an insulin-dependent manner. We demonstrate that this novel domain is relatively small (ϳ50 amino acids) and is located between the PH and SH2 domains. We therefore term it the BPS domain to denote this location. We present data to suggest that the two binding domains within Grb10 may play a role in determining the specificity of receptor interaction with Grb10.

MATERIALS AND METHODS
Yeast Strains and Plasmids-S. cerevisiae EGY48 (a-trp1, ura3-52, his 3, leu2) and all yeast expression plasmids were provided by the laboratory of Roger Brent and have been previously described (3,4,(33)(34)(35). All procedures for routine growth and maintenance of yeast strains were described previously (36). Plasmid transformation of yeast cells was by the lithium acetate method (37). The insulin receptor and IGFI receptor cDNA fusions have been previously reported (3,4,15).
Plasmid Construction-Full-length and various subdomains of the human Grb10 cDNA were subcloned into the two-hybrid plasmid pJG4 -5 using standard methods (38). The plasmids encoding the wildtype, K1018A, and Y960F IR hybrids and the K1003A and Y950F IGFIR hybrids have been previously reported (4,15). All site-directed mutants were generated by the method of Kunkel (39). Detailed cloning strategies for all constructs are available upon request.
␤-Galactosidase Assay-The colony color and solution ␤-galactosidase assays were performed as described (4,40), and the units of ␤-galactosidase activity were calculated by the method of Miller (41). The values shown represent averages of more than three assays Ϯ S.E. (each assay representing an independent colony).
Microinjection Studies-NWTb3 cells, which overexpress the IGFIR, were maintained as described previously (19) and grown on glass coverslips for microinjection experiments. The cells were rendered quiescent by serum deprivation for 24 -36 h prior to injection. GST-Grb10 fusion proteins were expressed in DH5␣ cells and purified with glutathione-agarose beads by standard techniques (38). The fusion proteins were microinjected at a concentration of 3.5 mg/ml, and preimmune sheep IgG was added to the protein for detection purposes. Parallel injections of sheep IgG alone were also performed. One hour after injection, the cells were stimulated as indicated with 100 ng/ml IGFI. A solution of 3-bromo-5Ј-deoxyuridine (BrdUrd) (Amersham Pharmacia Biotech) was also added to the medium for 16 h to label newly synthesized DNA. Following the labeling period, the cells were fixed and processed for immunofluorescence as described by Xiao et al. (9). Briefly, the cells were stained with a rat monoclonal anti-BrdUrd antibody (Accurate Scientific), followed by a tetramethylrhodamine B isothiocyanate-conjugated donkey anti-rabbit IgG antibody and a fluorescein isothiocyanate-conjugated donkey anti-sheep IgG antibody to detect the injected cells. This technique results in red nuclear staining in cells that have progressed to S phase of the cell cycle and green cytoplasmic staining in the injected cells. Results were analyzed and quantitated using an epifluorescence photomicroscope (Carl Zeiss, Inc.).

Grb10 Interacts with the IR via the C-terminal SH2 Domain and a Novel Domain Located between the PH and SH2
Domains-Grb10 has been shown to interact directly with the IR and IGFIR (23-25, 30 -32). We had previously reported that the IR interacts strongly with a region of Grb10 that spans amino acids 358 -536, a region that includes the SH2 domain. Since this interaction was kinase-dependent in the two-hybrid assay and because in vitro interaction of Grb10-GST fusions with cell-derived receptors was insulin-dependent (23), we concluded that the interaction was mediated via the SH2 domain. In fact, it has been demonstrated that the SH2 domain of Grb10 does mediate an interaction with the IR (24, 25, 30 -32). However, in subsequent experiments we found that mutagenesis of the critical arginine (to lysine) within the SH2 FLVRES motif, which would be predicted to disrupt phosphotyrosine interactions, had only minor effects upon full-length Grb10 interaction with the IR despite the demonstration that the mutant SH2 domain showed no interaction by itself. This prompted us to conclude that an additional interaction domain must exist within Grb10. To delineate this binding domain, we again utilized the yeast two-hybrid system to test the interaction of a number of Grb10 fragments with the IR. As shown in Fig. 1, full-length Grb10 interacted well with the IR as did the full-length Grb10, which contained a single amino acid substitution in the SH2 domain, in which Arg 462 within the conserved FLVRES motif was substituted with Lys. This substitution is known to cause a dramatically reduced interaction with phosphotyrosine-containing proteins in a variety of SH2 domains. To ensure that this mutation would indeed reduce the interaction of the Grb10 SH2 domain with the IR, we expressed a fusion protein that consisted of only the mutated SH2 domain (SH2-R/K, amino acids 413-536). As expected, this mutated SH2 domain showed no interaction with the IR, despite equal expression levels compared with the intact SH2 domain (data not shown). To localize the additional interaction domain we expressed a number of regions of Grb10 as fusion proteins and assayed their interaction with the IR. As shown in Fig. 1, the amino terminus of Grb10, up to and including the PH domain, showed no interaction with the IR; nor did the PH domain by itself. As expected, all constructs that contained an intact SH2 domain showed high activities in the two-hybrid assay, showing clearly that the SH2 domain can interact independently with the IR. An important clue to the location of the additional binding domain came when we expressed the 358 -536 region of Grb10, which contained the mutated SH2 domain. This hybrid protein showed an activity in the two-hybrid assay that was equivalent to its wild-type counterpart. This result suggested that the additional domain was either immediately N-terminal to the SH2 domain or overlapped the SH2 domain. To test this idea, we expressed the region of Grb10 between the PH and SH2 domains (termed BPS) and observed a high level of activity in the two-hybrid assay. Consistent with our previous data (23), none of these interactive Grb10 regions showed any interaction with a kinase-inactive IR (data not shown).
The Domain between the PH and SH2 Domains (BPS) Inter-acts Independently with the IR-As shown schematically in Fig. 2, the known members of the Grb7/Grb10/Grb14 protein family show a high degree of identity within the SH2 domain and the conserved 300-amino acid central domain (26,29), which includes the PH domain, and two conserved blocks of ϳ100 amino acids each located both N-terminal and C-terminal to the PH domain. We term the domain between the PH and SH2 domains BPS. To further examine the receptor binding activity of the BPS region, we generated a series of hybrids from this region and assayed their interaction with the IR in the two-hybrid system. As shown in Fig. 2, interaction of BPS domain amino acids 358 -434 resulted in high activity, which was approximately equivalent to that observed with the SH2 domain. Further truncation of the C terminus of the BPS region to amino acid 412 or 402 showed no apparent loss of activity. Truncation of the N-terminal residues to 364 showed no significant loss of activity, whereas further truncation to residue 369 showed consistently reduced activity. The kinaseinactive IR (K1018A) hybrid showed no interaction with any BPS domain hybrid proteins (data not shown). We conclude that a distinct domain exists between the PH and SH2 domains, which is capable of binding to the IR in a manner independent of the SH2 domain. Furthermore, the interaction depends upon IR kinase activity. A BLAST search of the data bases using the amino acid sequences between the PH and SH2 domains showed that only the Grb7/10/14 proteins gave significant scores (p ϭ 1.2e Ϫ16 or lower; data not shown), suggesting that this domain is not found in other known proteins.
Both the SH2 and BPS Domains of Grb10 Interact with the IR in Vitro-To further examine the SH2 and BPS domain interactions, we produced nine GST-Grb10 fusion proteins corresponding to the PH, SH2, and BPS domains (Fig. 3A). Immobilized GST fusions were incubated with lysates derived from CHO-IR cells that were or were not insulin-stimulated. After extensive washing, the precipitated proteins were resolved by SDS-polyacrylamide gel electrophoresis and subsequently immunoblotted with either anti-phosphotyrosine or anti-IR antibodies. Neither the GST protein alone nor the GST-PH fusion showed any in vitro interaction with the IR (Fig. 3B, GST and lane 3). As we observed previously, the Grb10 hybrid containing both the BPS and SH2 domains interacted strongly with the IR (23). Consistent with our two-hybrid data, mutation of R462K within the Grb10 SH2 FLVRES motif had no significant effect on interaction of the BPS plus SH2-containing fusion (compare lanes 1 and 2). The SH2 domain alone interacted efficiently with the IR (lane 4), and as expected, the R462K mutation eliminated all SH2 binding (lane 5). Several GST fusions containing various BPS domain residues (amino acids 358 -434, 364 -412, and 369 -412) were also found to interact efficiently with the IR to an extent roughly similar to that of the SH2 domain (lanes 6 -8), whereas the BPS amino acids 369 -402 bound only very weakly to the IR (lane 9). The majority of these results are consistent with our findings from the yeast two-hybrid assay. However, the 369 -402 fusion showed less in vitro binding activity (Fig. 3B, lane 9) than was expected from the two-hybrid assay. Consistent with this finding, two additional GST constructs, 358 -402 and 364 -402, both showed poor interaction in vitro (data not shown), suggesting that amino acids between 402 and 412 are important for in vitro interaction but are not necessary for the more sensitive two-hybrid assay. Alternatively, these hybrid proteins may not fold properly in bacteria as compared with yeast. To prove that the SH2 and BPS domains interact specifically with the activated IR and not the unstimulated IR, we immunoblotted the precipitates with an antibody against the IR (Fig. 3C). These fusion proteins only showed interaction with the IR derived FIG. 1. Grb10 interacts specifically with the IR via the SH2 domain and a second region located between the PH and SH2 domains. Various regions of Grb10 were assayed for interaction with the wild-type IR "bait" hybrid in the yeast two-hybrid assay. Transformants were assayed for ␤-galactosidase activity by the solution assay as described under "Materials and Methods." ␤-Galactosidase (␤-Gal) activity is reported in Miller units, and the data represent an average from a minimum of three independent colonies. ϩϩϩ, a dark blue color in the colony color assay; Ϫ, white colonies.

Dual Receptor Interaction Domains within Grb10
from insulin-stimulated CHO/IR cells, showing that both the SH2 and BPS bind in an insulin-dependent manner.
The Grb10 BPS Domain Interacts More Strongly with the IGFIR than the SH2 Domain-The IGFIR shares a high sequence homology with the IR (45). Nevertheless, coimmunoprecipitation studies of the Grb10 protein with either the IR or IGFIR have suggested that Grb10 associates preferentially with the IR (30). Thus, we investigated the Grb10 interaction with the IGFIR to see if these results might be attributed to differential Grb10 domain interaction. We therefore examined the interaction of Grb10 with the IGFIR using GST pull-down assays. As shown in Fig. 4, the control GST and PH domain fusions did not precipitate any IGFIR. Surprisingly, although the Grb10 SH2 domain fusion showed significant activity with the IGFIR in the yeast two-hybrid assay (ϳ80 -100 units, data not shown), the GST-SH2 hybrid showed virtually no interaction with IGFIR in vitro. In comparison, the GST-BPS fusion containing amino acids 364 -412 interacted efficiently with the IGFIR in vitro. Interestingly, despite the apparent lack of interaction of the SH2 domain and the IGFIR in vitro, the BPS plus SH2 fusion protein consistently coprecipitated more IG-FIR than the BPS domain alone. In addition, the R462K mutation introduced into the BPS plus SH2 construct reduced the interaction by about 50%, yet this mutant protein still showed a relatively stronger interaction with the IGFIR than the BPS domain alone. These data suggest that the SH2 and BPS domains may bind to the IR and IGFIR in a cooperative manner and that even the mutant SH2 domain may stabilize and strengthen the BPS domain interaction as long as the BPS domain is present to direct the SH2 domain to the receptor. Alternatively, the SH2 domain may act to stabilize the struc-ture of the BPS domain such that it binds more efficiently to the IR. These data with the mutated SH2 domain are in agreement with results from the yeast two-hybrid assay, since the 358 -536 (BPS plus SH2) showed similar activities whether or not they contained the SH2 mutation (Fig. 1).
The BPS Domain Does Not Interact with the EGFR-Although the murine Grb10 protein was initially discovered by screening an expression library with a 32 P-labeled cytoplasmic domain of the EGFR, in vivo interaction has not been easily demonstrated (26). Furthermore, it has been shown that GST-Grb10 fusion proteins interact with the EGFR with less avidity than with the IR and platelet-derived growth factor receptor (24,25). We reasoned that the weak interaction with the EGFR might be due to differential association of the receptor with the two domains of Grb10. To address this question, we examined EGFR coprecipitation with the previously described Grb10 GST fusions. As shown in Fig. 4, only the SH2 domain was able to bind efficiently to the EGFR in vitro. The BPS domain (364 -412) showed no binding to the EGFR. In support of this, the substitution mutation (R462K) within the Grb10 SH2 domain totally abolished this interaction with both the SH2 and the BPS plus SH2 hybrids.
IR Domains Required for Interaction with the BPS and SH2 Domains of the Grb10 -Several tyrosine residues in the cytoplasmic domain of the IR and IGFIR become phosphorylated during receptor autophosphorylation, and these phosphotyrosines serve either to activate the kinase activity of the receptor (2) or act as docking sites for substrates of receptors. Specifically, the three tyrosines (positions 1146, 1150, and 1151) within the kinase domain activation loop play a role in activation of the IR catalytic activity (46). In addition, two juxtamem-

FIG. 2. Grb10 contains a second conserved domain between amino acids 364 and 412, which interacts independently with the IR.
Top, sequence alignment between the BPS domain of Grb10 and the homologous regions of other members of the Grb10-related proteins Grb7 and Grb14. The numbers refer to percentage of amino acid identity. Bottom, six Grb10 hybrid proteins were produced as activation domain fusions and tested for interaction with the IR in the two-hybrid assay as described under "Materials and Methods." brane tyrosines become phosphorylated (953 and 960), the latter of which is located within an NPXpY motif (where pY represents phosphorylated tyrosine), which is required for binding to the phosphotyrosine binding domains within IRS-1, IRS-2, and SHC (3)(4)(5). In addition, two tyrosines within the C terminus become phosphorylated (1316 and 1322), and although 1322 has been shown to interact with p85 (47), the role of these tyrosines remains unclear. The sites of interaction of Grb10 with the IR has been controversial, with one group suggesting that the C terminus of the IR was the primary binding site (31). Another group has presented data supporting a role for phosphotyrosines within the juxtamembrane domain and within the kinase domain (25). To further address this issue, we utilized the yeast two-hybrid system in an attempt to further define the binding sites for the SH2 and BPS domains of Grb10 within the IR. We utilized a number of IR mutants and assayed their activity with the SH2 domain, BPS domain, and BPS plus SH2 domain fragments of Grb10. The SH2 domains of p85, previously shown to interact well with the IR (3), was used as a positive control. As shown in Fig. 5A, mutation of either of the juxtamembrane tyrosines, 953 or 960, had no effect upon interaction with a Grb10 protein containing both the SH2 and BPS domains. Likewise, deletion of the carboxylterminal 30 amino acids from the IR (which deletes the two C-terminal autophosphorylation sites) had no effect. Mutation of Y1146, the most N-terminal of the three tyrosines within the IR activation loop also had no effect. However, substitution of the other two activation loop tyrosines (1150/1151) with phenylalanine essentially eliminated interaction with the BPS plus SH2 construct. Importantly, this receptor mutant shows significant interaction with p85 (Fig. 5D), showing that this receptor is catalytically active, and suggests that the abolition of interaction of the IR with Grb10 may be due to elimination of a specific interaction site for Grb10 within the IR rather than being due to a global loss of catalytic activity by this IR mutant. We next examined the independent interaction of either the SH2 domain or the BPS domain with these same IR mutants. As shown in Fig. 5, B and C, both the SH2 and BPS domains showed markedly reduced interaction with the Y1150F/Y1151F double substitution mutant, showing that this receptor region is essential for interaction with both domains. It is of note that the p85 SH2 domain, which has been shown to interact with the C-terminal tyrosine 1322 (47), retains an interaction with the C-terminally deleted IR. This appears to be due to interaction with another region of the IR, which may include the tyrosines within the activation loop (data not shown). We conclude from these studies that the paired tyrosines (Tyr 1150 and Tyr 1151 ) within the activation loop are essential for binding to both the SH2 and BPS domains of Grb10.
To further examine whether the SH2 and BPS domains within Grb10 can interact in vitro with the kinase domain of the IR, we tested the ability of the GST fusions to interact with a purified IR kinase domain produced in an insect cell (Sf9) expression system. We utilized a purified preparation of baculovirally expressed IR kinase domain (a gift of Steve Hubbard), which had been activated in vitro and consisted primarily of a form of the kinase domain that was triply phosphorylated within the activation loop (data not shown). As shown in Fig.  5E, the BPS plus SH2-GST fusion interacted well with the IR kinase domain whether or not the SH2 domain contained an Arg or Lys within the FLVRES box. The GST control and PH fusion showed no interaction with the IR kinase domain. Individually, both the SH2 and BPS domains showed interaction with the IRK3P, but at lower levels as compared with the BPS plus SH2 constructs (data not shown). These data are consistent with our two-hybrid results and clearly show that Grb10 can interact with the kinase domain in the absence of both the juxtamembrane and C-terminal IR domains. In addition, these data show conclusively that the interaction between these proteins is direct, since both the IR and Grb10 proteins are highly purified.
Microinjection of the Grb10 Region Containing the SH2 and BPS Domains Inhibits Mitogenesis Induced by IGFI, while the  Fig. 3 were used to assay coprecipitation of the IGFIR using lysates prepared from NWTb3 cells, which overexpress the IGFIR. The lysates derived from hormone-stimulated or unstimulated cells were incubated with either a control GST or GST-Grb10 fusions, and the coprecipitating proteins were analyzed by blotting with an antiphosphotyrosine antibody Lower panel, the identical GST fusions were used to examine in vitro interaction with the EGFR expressed in 5HR11 cells. IB, immunoblot

FIG. 3. In vitro interaction of Grb10 SH2 and BPS domains with the insulin receptor.
A, we produced nine Grb10 GST-fusion proteins (1)(2)(3)(4)(5)(6)(7)(8)(9) corresponding to the PH domain, SH2 plus BPS domains, SH2 domain, and BPS domain of Grb10. B, these GST-Grb10 fusions were assayed for their ability to interact in vitro with the IR expressed in CHO cells. Cellular lysates were prepared from CHO cells, which overexpress the IR, stimulated with or without insulin (100 nM, 10 min), and incubated with either a control or GST-Grb10 fusion proteins as described under "Materials and Methods." The filters were blotted with antiphosphotyrosine antibodies. The numbers under the blots refer to the numbers of the GST fusions shown in A. C, similar experiments were performed with three of the above GST fusion proteins. The filter was blotted with an anti-IR antibody. IB, immunoblot.
Individual Domains Have No Effect-As a means of examining the effects of these Grb10 domains on IGFI signaling in a cellular system, we produced GST fusion proteins, which contained the Grb10 PH, SH2, and BPS domains. We have previously used this approach to examine the effect of a Grb10 fragment (amino acids 358 -536) upon insulin, IGFI, epidermal growth factor, and serum-stimulated mitogenesis (23). Injection of this fusion protein inhibited insulin-and IGFI-stimulated progression to S phase but had no effect upon epidermal growth factor-or serum-stimulated cell cycle progression. Since this Grb10 fusion protein contains both the SH2 and BPS domain (which we were previously unaware of), it was necessary to examine the effects of injecting each domain independently to examine its effect upon cell cycle progression. These new fusion proteins were microinjected into fibroblasts that overexpress the IGFIR, and the injected cells were monitored for mitogen-stimulated BrdUrd incorporation using previously described methods (19). In this assay, all microinjection samples contained preimmune IgG to allow the detection of injected cells, which were then scored as positive or negative for cell cycle progression based upon the presence or absence of nuclear rhodamine staining. As shown in Fig. 6, treatment of cells with IGFI resulted in an increase in the percentage of BrdUrdpositive cells from a basal level of 16.3% to an average of 56.1%. Microinjection of the PH domain had no effect upon BrdUrd incorporation (52.8%). Injection of the individual domains encoding the SH2 (49.5%), SH2-R/K (43.9%), or BPS (48.6%) domain also showed no significant inhibition of cell-cycle progression. Injection of either the BPS plus SH2 or the BPS plus SH2-R/K constructs did show significant inhibition of IGFI-stimulated mitogenesis of ϳ50% (32.2 and 33.1%, respectively) as compared with IgG-injected, IGFI-treated cells. These data suggest that efficient interaction of Grb10 with the IGFIR requires both the SH2 and BPS domains. Interestingly, as we had previously observed in the two-hybrid and GST pull-down assays, a BPS plus SH2 construct with a mutated SH2 FLVRES motif retained the ability to inhibit IGFIR mitogenic signaling, suggesting that the BPS and SH2 domains may bind in a cooperative manner and that the mutated SH2 domain may be able to interact with the IR but only if the BPS domain is also present. Alternatively, the mutated SH2 domain may act sterically to more effectively block interactions with endogenous signaling proteins.
Potential Roles of Multiple Receptor Interaction Domains within Grb10 -Our results clearly demonstrate the existence of a second domain within Grb10 that is capable of mediating a kinase-dependent interaction with the IR and IGFIR. This region, which we have termed BPS, is highly conserved within the related proteins Grb7 and Grb14, and it is likely that these domains may play a similar role within these proteins. Our data show that the BPS domain binding to the IR requires receptor kinase activity and more specifically appears to require the paired tyrosines within the activation loop within the kinase domain. What remains unclear is whether the BPS domain interacts directly with the activation loop tyrosines or alternatively whether autophosphorylation of these tyrosines causes a conformational change within the kinase domain, which exposes a binding site for the BPS domain. Recent structural studies in which the activated IR kinase domain has been crystallized have shown that the activation loop does in fact move a considerable distance after receptor autophosphorylation allowing unrestricted ATP and substrate access to the active site (46,48). In addition, the N-terminal lobe of the IR kinase domain shows a conformational change in which the ␣C helix shows an altered structure in the active state, thus exposing a number of hydrophobic residues that are buried in the inactive state (48). Thus, it is possible that the BPS domain may not interact directly with the paired phosphotyrosines but rather may interact with regions of the IR that undergo conformational alterations after IR autophosphorylation. In support of this, in preliminary studies we have been unable to demonstrate competition of in vitro interaction with the IR and the BPS GST hybrid using a variety of IR-derived phosphopeptides including activation loop peptides (data not shown). In this regard, given the sequence similarity of between the IR FIG. 5. Binding sites for the BPS and SH2 domains of the Grb10 within the IR. Various IR mutants were tested for interaction with three Grb10 hybrids containing the BPS plus SH2 domain (A), the SH2 domain (B), or the BPS domain (358 -412) (C), using the yeast twohybrid assay. D, a p85 domain hybrid containing both SH2 domains was used as a positive control. The numbers refer to percentage of activity compared with the wild-type (WT) IR control. E, baculovirally expressed, purified, and autophosphorylated IR kinase domain (IRK3P) was examined for in vitro interaction with purified Grb10-GST fusions. After incubation, the precipitates were washed, separated by SDSpolyacrylamide gel electrophoresis, and immunoblotted with antiphosphotyrosine antibodies (PY20). The control (Con) lane represents autophosphorylated IRK3P.
FIG. 6. Effects of microinjection of Grb10 GST fusions upon IGFI-mediated mitogenesis. Serum-starved IGFIR-overexpressing cells were injected with a GST-GRIP2 fusion proteins at a concentration of 3.5 mg/ml containing sheep IgG (5 mg/ml) for detection purposes. Parallel injections of sheep IgG alone were also performed. The cells were then stimulated as indicated with IGFI (100 ng/ml) in the presence of BrdUrd. DNA synthesis was quantitated following immunofluorescence staining for injected IgG as described under "Materials and Methods." Shown are the results obtained from two identical experiments in which at least 200 cells were injected on each coverslip. Bars, the mean result Ϯ S.E. and IGFIR in the activation loop, it is surprising that the SH2 domain appears to interact preferentially with the IR in vitro (Fig. 4) and perhaps in vivo (30). It remains possible that these two receptors may have distinct structures in the active state that confer differential binding profiles to receptor substrates.
Our demonstration of two independent interaction domains within a single signaling protein such as Grb10 is interesting for a number of reasons. First, as we have discussed, this may be a mechanism that helps to determine the specificity by which signaling proteins such as Grb10 interact with different receptor tyrosine kinases or other phosphoproteins. As shown schematically in Fig. 7, our data suggest that Grb10 may interact strongly with the IR by virtue of its interaction with both the SH2 and BPS domains. Conversely, Grb10 may interact less efficiently with the IGFIR and EGFR due to preferential interactions with either the BPS or the SH2 domain, respectively. In this regard, it is becoming increasingly clear that signaling proteins that contain multiple SH2 domains show a much higher affinity of interaction with their bis-phosphorylated peptide binding partners if both SH2 domains are expressed together (49,50). In other words, two binding domains within a molecule have been shown to lead to greatly increased affinity, which has been shown to be increased by many orders of magnitude as compared with the individual binding domains. Thus, it is possible that the reason that Grb10 can be coimmunoprecipitated efficiently with the IR but not with the IGFIR or EGFR may be explained by these findings (25,26,30). Thus far, we have only considered the possibility that Grb10 may be interacting with a single subunit of the IR. It is also possible that the two binding domains within Grb10 may facilitate interactions between IR cytoplasmic subunits, either between different monomers of the holoreceptor or even between distinct receptors.
Our findings may also cast some light upon the functional studies that have previously been described. It has been shown, for example, that overexpression of Grb10␣ in a CHO cell line resulted in decreased IRS-1 and p60 phosphorylation and reduced phosphatidylinositol 3-kinase activity (24). It has also been shown that microinjection of a fusion protein containing the BPS and SH2 domain inhibits insulin-mediated mitogenic signaling in rat1 fibroblasts overexpressing the IR (23). Given our findings suggesting that Grb10 may interact within or near the activation loop within the IR kinase domain, it seems possible that Grb10 may physically block access of substrates or ATP to the catalytic pocket, which is located very close to the activation loop. However, we cannot rule out the idea that Grb10 might also serve to increase signaling by the IR perhaps by stabilizing the phosphorylated activation loop in the activated configuration or by protection of these phosphotyrosines from tyrosine phosphatases. Further experiments will be required to address these issues. A real understanding of the function of Grb10 in receptor tyrosine kinase signaling is likely to become clear only when we begin to identify components that lie downstream of Grb10. FIG. 7. Model of Grb10 interaction with the IR, the IGFIR, and the EGFR. Our data suggest that both the SH2 domain and the BPS domain interact strongly with the IR. The BPS domain interacts well with the IGFIR, whereas the SH2 domain binds weakly as compared with the IR. The SH2 domain interacts well with the EGFR, while the BPS domain shows no interaction with the EGFR. This differential interaction may result in relatively high affinity binding of Grb10 to the IR as compared with the IGFIR and EGFR.