Activation of the Eck Receptor Protein Tyrosine Kinase Stimulates Phosphatidylinositol 3-Kinase Activity*

The Eph/Eck subfamily of receptor protein tyrosine kinases is currently the largest subfamily of receptor protein tyrosine kinases with a dozen members and Rev. Cell Biol. 10,251-337). Using the cytoplasmic domain of Eck as bait in a yeast two-hybrid screen of mouse embryonic and T-cell cDNAlibraries, it was discovered that the p85 subunit of phosphatidylinositol 3-kinase bound Eck. Further, using glutathione S-transferase fusion proteins, it was found that the C-terminal src homology 2 domain of the p85 subunit specifically interacted with Eck. Additionally, Eck coimmunoprecipitated with p85 in ligand activated cells confirming their interaction in vivo. In keeping with the above observations, activation of Eck by its ligand, B61, increased phosphatidylinositol 3-kinase activity. This is the first description of a signal transduction pathway initiated by any member of the EphlEck family. anti-phospho-tyrosine monoclonal antibody (UBI), followed by incubation with horse- radish peroxidase-conjugated goat anti-mouse IgG (Bio-Rad). After extensive washing, membranes were developed using a chemiluminescent reaction (ECL, Amersham Corp.) according to the manufacturer's in-structions. Reprobing of the membrane to detect Eck protein was done by incubating the above blot in TBS-T containing 1% BSA and 0.02% sodium azide overnight, followed by extensive washing with TBS-T and then incubation with 1 pg/ml of anti-Eck antibody for 1 h at room temperature. PI 3-Kinase Assays-PI 3-kinase on anti-phosphotyrosine (4G10) immunoprecipitates

A recently described powerful technique for the detection of protein-protein interactions in vivo is the yeast two-hybrid system (6). The interaction of a protein of interest expressed as bait with that encoded by a transfected library plasmid (prey) results in the activation of marker genes (such as P-galactosidase) that allows for easy detection. To identify molecules that bind the cytoplasmic domain of Eck upon activation, we used it as bait in a yeast two-hybrid screen. It was assumed that the cytoplasmic domain would autophosphorylate itself (due to its intrinsic tyrosine kinase activity) as this is a prerequisite for interacting molecules to bind through their SH2 domains that specifically recognize phosphotyrosines (7). Support for such an assumption came from the finding that Src, a cytoplasmic tyrosine kinase, does undergo autophosphorylation when expressed in yeast (8) and that the cytoplasmic domain of Eck expressed either by in vitro translation or following transfection of 293T cells possesses potent tyrosine kinase activity (data not shown).
Two interacting clones were isolated that encoded different regions of the @ isoform of the p85 subunit of PI 3-kinase. PI 3-kinase is a heterodimer composed of a regulatory p85 subunit (a and @ isoforms have been identified) and a p l l 0 catalytic subunit (9). Binding of ligand to a multitude of receptors including the platelet-derived growth factor receptor (PDGFR), epidermal growth factor receptor (EGFR), interleukin (IL)-2 (IL-2), IL-4, and IL-7 receptors stimulates PI 3-kinase activity which phosphorylates the head group of phosphatidylinositol (4,5)-bisphosphate (PtdIns(4,5)P2,) at the D-3 position to yield a highly polar membrane lipid PtdIns(3,4,5)P3 which has been postulated to act as a second messenger (10). The downstream intracellular targets of PI 3-kinase are not known, though a recent study indicated that PtdIns(3,4,5)P3 could activate a phorbol 12-myristate 13-acetate-insensitive isoform of protein kinase Cg (11). Other evidence implicating PI 3-kinase in signal transduction includes the finding that the transforming activity of the polyoma virus middle T antigen is dependent on the binding of PI 3-kinase to pp60'"" and that activation of PI 3-kinase is required for platelet-derived growth factor-@ receptor internalization (12,13).
The p85 subunit is a modular adapter that contains an SH3 domain (binds proline-rich sequences), two SH2 domains (specific for phosphotyrosines in a linear pYXXM motif) and a unique N-terminal region (1,10,14). Either or both of the SH2 domains (N-terminal (N-SHB), C-terminal (C-SH2)) could POtentially mediate binding to the phosphorylated cytoplasmic domain. However, it was found that almost exclusively, the C-SH2 domain bound to the Eck cytoplasmic domain. Confirming this was the finding that the C-SH2 domain expressed as a GST fusion protein precipitated the Eck RPTK from vascular smooth muscle cell lysates. Furthermore, anti-p85 antibody coimmunoprecipitated Eck from B61-stimulated cells indicating that Eck interacts with PI 3-kinase in viuo. Finally, activa-tion of the Eck RF'TK by its cognate ligand B61 resulted in the stimulation of PI 3-kinase activity.

MATERIALS AND METHODS
Yeast no-hybrid Screen and cDNA Isolation -The cytoplasmic domain of the human Eck RPTK was obtained by polymerase chain reaction (PCR) using a full-length Eck cDNA (2) as template and custom oligonucleotide primers. The sense primer, including a custom SfiI site :underlined) had the sequence:

CTGGCCATGGAGGCCCACCGCAG-
.AGGAAGAACCAG and the antisense primer including a custom Sal1 site had the sequence:

CGACTGTCGACTCAGATGGGGATCCCCA-
CAGTGTTCAC. The amplified fragment was cloned into the yeast bait plasmid pASlCYH2 to be expressed as a hybrid gene consisting of an upstream hemagglutinin epitope tag and downstream the DNA-binding domain of GAL4 (15). This bait plasmid was cotransformed with a mouse T-cell expression library fused to the GALA activation domain in the PACT plasmid (prey) or with a day 10.5 mouse embryonic expression library fused similarly to the VP16 activation domain (16, 17).
Several of the lo6 transformants screened had detectable P-galactosidase activity. Library plasmid recovered from the positive clones was used in a cotransformation assay with either the Eck cytoplasmic bait or other control heterologous baits. Six plasmids were isolated that encoded proteins that specifically interacted with Eck. DNA sequencing and data base searching revealed two of these plasmids to encode regions of the p85 subunit of PI 3-kinase.
I n Vitro Dunscription and Dunslation-The p85 subunit of PI 3-kinase was amplified by PCR using an upstream primer that included a custom T7 promoter site (lower case) and library plasmid (PACT) sequence representing the GAL4 activation domain: taatacgactcactatagG-GAGACCACATGGATGATGTATATAACTATCATCATTTC. The downstream primer corresponded to sequences in the ADH terminator of the library plasmid: CTACCAGAA'M'CGGCATGCCGGTAGAGGTGTGGTCA. This amplified fragment was used for coupled in vitro transcription and translation using the TNT T7 coupled reticulocyte lysate system (Promega, Madison, WI) and [3sS]methionine (Amersham Corp.) according to the manufacturers' instructions.
Production of GST Fusion Proteins-The Eck RPTK cytoplasmic domain (amino acids 559-976) was obtained by PCR employing Eck cDNA as template and an upstream primer containing a custom EcoRI site: GGAATTCACCGCAGGAGGAAGAACCAG and a downstream primer containing a custom Sal1 site: CGACTGTCGACTCAGATGGGGATC-CCCACAGTGTTCAC. To obtain the deleted form of the Eck RPTK cytoplasmic domain, AEck (amino acids 559-642), a different downstream primer with a custom Sal1 site (underlined) and an artificial in-frame stop codon (italics) was used for PCR amplification: ATAG-CAGTCGACTCACGGCACCTCCTTCTTCC. The fragments were subcloned into the GST fusion protein vector pGSTag (18) and transformed into the E. coli strain BL21fDE3) pLysS. GST and GST fusion proteins were prepared using published procedures (19) and the recombinant proteins immobilized onto glutathione-agarose beads (Sigma). GST fusions of N-SH2 and C-SH2 domains of the p85 subunit of PI 3-kinase were prepared as described previously (20L2 GSTBinding Assays-10 p1 of ["Slmethionine-labeled in vitro translated material or an equal number of trichloroacetic acid precipitable counts from lysates of 1 x lo6 metabolically labeled cells were incubated with equivalent amounts of GST fusion proteins in 1 ml of GST-binding buffer (50 mM Tris, pH 7.6, 150 mM NaCI, 2 mM EDTA, 0.5% Triton X-100, 10% glycerol) in the presence of protease inhibitors (5 pg/ml leupeptin, 5 pg/ml aprotinin, 50 pg/ml soybean trypsin inhibitor, and 5 pg/ml pepstatin) and 1 mM sodium orthovanadate for 1 h a t room temperature. Following incubation, beads were washed three times in GSTbinding buffer, boiled in SDS sample buffer (containing 2% P-mercaptoethanol), and resolved on a 10% SDS-polyacrylamide gel. Bound proteins were visualized by autoradiography.
Metabolic Labeling and Immunoprecipitations-Rat vascular smooth muscle cells (SMCs) were isolated and passaged as described previously (21). They were brought to quiescence by maintaining in serum-free medium (Eagle's minimal essential medium containing 1% bovine serum albumin (BSA)) for 48 h and then metabolically labeled with 100 pCi/ml of  GST only, AEck GST or Eck GST bound to agarose beads and processed as in A. 0.5% Triton X-100, 10% glycerol and 1 mM sodium orthovanadate in the presence of protease inhibitors. Cleared cell lysates were incubated with either control rabbit serum, anti-Eck antibody (2) or the indicated GST fusion proteins for 2 h a t 4 "C. To show interaction of Eck with p85 subunit, cells were either untreated or treated with 1 pg/ml of recombinant B61 (5) for 5 min. Immunoprecipitation with anti-rat p85 polyclonal antibody (UBI, Lake Placid, New York) was done as described previously (20).
Lysates incubated with GST fusion proteins were similarly washed and bound material eluted by boiling in 1% SDS. The eluates were then diluted ten-fold with lysis buffer and immunoprecipitated with anti-Eck antibody. For immunoblotting with the antiphosphotyrosine antibody, cells were starved as above, treated with 1 pg/ml B61-Ig or control-Ig chimera (22) and then lysed in lysis buffer containing 50 mM Tris, pH 7.6, 150 mM NaCI, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, 1 mM sodium orthovanadate, and protease inhibitors. The cell lysates were incubated with 5 pg/ml of anti-Eck antibody and processed as described below.
Western Blot Analysis-Precipitated immune complexes were resolved by SDS-PAGE, transferred to nitrocellulose by electroblotting, blocked with 1% BSA in Tris-buffered saline containing 0.1% Tween (TBS-T) overnight a t 4 "C and then incubated with 4G10 anti-phosphotyrosine monoclonal antibody (UBI), followed by incubation with horseradish peroxidase-conjugated goat anti-mouse IgG (Bio-Rad). After extensive washing, membranes were developed using a chemiluminescent reaction (ECL, Amersham Corp.) according to the manufacturer's instructions. Reprobing of the membrane to detect Eck protein was done by incubating the above blot in TBS-T containing 1% BSA and 0.02% sodium azide overnight, followed by extensive washing with TBS-T and then incubation with 1 pg/ml of anti-Eck antibody for 1 h at room temperature.

RESULTS AND DISCUSSION
The cytoplasmic domain (amino acids 559-976) of the human E c k RPTK, which when expressed in 293T cells displayed constitutive tyrosine kinase activity (data not shown), was fused in f r a m e t o the GAL4 DNA-binding domain in the yeast vector pASlCYH2. This was used as bait to detect interacting proteins encoded by library cDNAs fused to the GATA or VP16 activation domains that were generated from mouse T-cells or mouse 10.5 day embryonic tissue, respectively (17,19  transformants were screened by expression in a yeast strain harboring lac2 and HIS3 genes under control of the GAL4 upstream activating sequence. Only transformants in which HIS3 is activated will grow in the presence of 3-amino-1,2,4triazole, an inhibitor of the HIS3 gene product. These colonies were in turn screened for lac2 expression. Library plasmids recovered from such positive colonies were then subjected to another round of cotransformation with either Eck or other heterologous baits. In this manner, library plasmids that specifically interacted with Eck were identified and marked for further characterization. One interacting clone from the embryonic cDNA library encoded only the C-terminal SH2 domain of the /3 isoform of the p85 subunit of PI 3-kinase (amino acids 579-724), while another from the T-cell library encoded both the N-and C-terminal SH2 domains of the / 3 isoform of the p85 subunit (amino acids 335-724). To obtain independent confirmation of the interaction, a cDNA encoding the cytoplasmic domain of the Eck RPTK was in vitro transcribed and translated using a rabbit reticulocyte lysate system and a protein of the predicted molecular weight (46 kDa) obtained that specifically bound the C-SH2 domain of the p85 subunit expressed as a GST fusion protein (Fig. IA). No detectable binding was observed with GST alone or with the N-SH2 domain expressed as a GST fusion protein despite the ability of the N-SH2 GST to precipitate EGFR and PDGFR (2O).To further validate this finding, the same experiment was performed in reverse. To characterize this interaction in vivo, it was first important to identify an easily manipulatable primary cell strain that expressed the Eck RPTK. A number of primary cell cultures were screened by flow cytometry and rat vascular SMCs were found to express significant amounts of Eck (data not shown). To confirm the presence of Eck, the SMCs were metabolically labeled and subjected to immunoprecipitation with an anti-Eck antibody. As shown in Fig. 2A, a protein of the expected molecular mass of 130 kDa was specifically precipitated by the anti-Eck antibody but not by control serum. In addition, the Eck ligand B61 expressed as an Ig-chimera (B61-121, but not a control-Ig (22) chimera, induced autophosphorylation of Eck in these cells (Fig. 2B, top panel). The bottom panel shows the same blot reprobed with anti-Eck antibody.
To confirm the interaction of Eck with p85 and to delineate which SH2 domain of p85 was capable of binding Eck, metabolically labeled rat SMC lysates stimulated with B61-Ig were precipitated with either the N-SH2 or C-SH2 domains of p85 expressed as GST fusion proteins2 (20). Bound material was eluted with 1% SDS, diluted, and reimmunoprecipitated with anti-Eck antibody. As shown in Fig. 3 A , Eck was specifically precipitated by C-SH2 GST but not by N-SH2 GST or GST alone. This finding is in agreement with other in vitro binding studies that indicate that high affinity binding in solution is entirely dependent on the C-SH2 domain (23). However, i n vivo, both SH2 domains may be required for stable binding (24). To prove that the Eck RPTK associates with p85 in vivo, lysates from unactivated or activated SMCs were immunoprecipitated with anti-p85 antibody prior to immunoblotting with anti-Eck antibody. As shown in Fig. 3B, addition of B61 induced the association of p85 with the Eck RF'TK.
To determine if activation of the Eck RPTK by its cognate ligand resulted in the stimulation of PI 3-kinase activity, quiescent SMCs were treated for 5 or 10 min with B61-Ig and cell lysates immunoprecipitated with an antiphosphotyrosine antibody to capture all tyrosine phosphorylated proteins. PI 3-kinase activity was then measured in the precipitated immune complexes and found to be increased %fold over basal activity within 5 min of activation and 4-fold over basal level by 10 min (Fig. 4). with B61-Ig for 5 or 10 min or with 10% fetal bovine serum for 5 min.
Taken together, these data indicate that the Eck RPTK is capable of binding to the C-terminal SH2 domain of the p85 subunit of PI 3-kinase. Furthermore, a deletion mutant of Eck (AEck) that removed the entire catalytic domain was incapable of binding to the p85 subunit. This is consistent with the removal of two YXXM motifs in this deletion mutant as this motif has previously been shown to mediate p85 subunit binding (25).
Finally and most importantly, the activation of Eck by its ligand B61 stimulated PI 3-kinase activity. Recently, we have found that activation of Eck by B61 results in ~hemotaxis.~ The relevance of this may lie in the finding that activation of PI 3-kinase is required for platlet-derived growth factor-induced membrane ruffling and chemotaxis (26) and that D-3 phosphorylated phosphoinositides may be involved in cytoskeletal alterations (27). Consistent with this is the observation that the p85 subunit has GTPase-activating protein activity toward Rac, which is intimately involved in membrane ruffling (28). It is therefore tempting to speculate that a similar role is being fulfilled by Eck-induced stimulation of PI 3-kinase.