Signaling inositol polyphosphate-5-phosphatase. Characterization of activity and effect of GRB2 association.

An inositol polyphosphate-5-phosphatase (SIP-110) that binds the SH3 domains of the adaptor protein GRB2 was produced in Sf9 cells and characterized. SIP-110 binds to GRB2 in vitro with a stoichiometry of 1 mol of GRB2/0.7 mol of SIP-110. GRB2 binding does not affect enzyme activity implying that GRB2 serves mainly to localize SIP-110 within cells. SIP-110 hydrolyses inositol (Ins)(1,3,4,5)P4 to Ins(1,3,4)P3. The enzyme does not hydrolyze Ins(1,4,5)P3 that is a substrate for previously described 5-phosphatases nor does it hydrolyze phosphatidylinositol (PtdIns)(4,5)P2. SIP-110 also hydrolyzed PtdIns(3,4,5)P3 to PtdIns(3,4)P2 as did recombinant forms of two other 5-phosphatases designated as inositol polyphosphate-5- phosphatase II, and OCRL (the protein that is mutated in oculocerebrorenal syndrome). The inositol polyphosphate-5-phosphatase enzyme family now is represented by at least 9 distinct genes and includes enzymes that fall into 4 subfamilies based on their activities toward various 5-phosphatase substrates.

There are now at least 9 distinct genes for 5-phosphatases or proteins with conserved 5-phosphatase sequences. The most recently identified of these are the 110-kDa SIP-110 (20) and the 133-kDa SHIP or SIP-130 (20 -24). cDNAs encoding these proteins were cloned from human and mouse cDNA libraries, respectively, based on their ability to associate with the adaptor protein GRB2 (20 -22) or their homology to 5-phosphatases or their ability to form complexes with the immunoreceptorbased tyrosine activation motif of mast cells (23,24). The predicted amino acid sequences of the human 110-kDa protein and the 133-kDa mouse form are 86% identical over 969 amino acids, and peptide sequence from the mouse 133-kDa protein suggests that they are alternatively spliced products of a single gene (20,21). Both associate with the SH3 domains of GRB2 through proline-rich sequences in their C-terminal portion. In addition, the 133-kDa protein contains an N-terminal SH2 domain that the 110-kDa protein lacks, becomes phosphorylated on tyrosine, and associates with the tyrosine-phosphorylated adaptor protein Shc (20 -22). These 5-phosphatases hydrolyze only the 3-phosphate-containing inositol phosphates, Ins(1,3,4,5)P 4 and PtdIns(3,4,5)P 3 (20,21,23). We now report the enzyme activity and products of recombinant 110-kDa SIP-110 and the effects of GRB2 on its activity.

EXPERIMENTAL PROCEDURES
Materials-All 3 H-labeled inositol phosphate isomers and [ 3 H]PtdIns(4,5)P 2 were purchased from DuPont NEN. PtdIns(4,5)P 2 and anti-HA antibody 12CA5 were purchased from Boehringer Mannheim. All unlabeled inositol phosphate isomers were purchased from Boehringer Mannheim or Calbiochem. Horseradish peroxidase-linked anti-mouse IgG was purchased from Bio-Rad. ECL Western blotting detection reagents were purchased from Amersham Life Sciences, and the 9E10 anti-Myc antibody was from Oncogene Science. Silica Gel 60 TLC plates (20 ϫ 20 cm, 0.2 mm) were from Merck. The Adsorbosphere SAX HPLC column was purchased from Alltech; the Partisil 10 SAX HPLC column was purchased from Whatman.
Expression of Recombinant Proteins-The complete coding sequence of SIP-110 (20) was expressed in Sf9 insect cells as a GST fusion protein with an intermediate HA tag using a baculovirus expression vector pVIKS as described (25). Human 5-phosphatase I was expressed as a 412-amino acid protein (26,27) using the pVL1392 baculoviral expression vector and BaculoGold transfection kit from PharMingen. Human 5-phosphatase II used in these studies was 5PtaseS consisting of amino acids 250 -942 of the predicted amino acid sequence (28). An N-terminal truncated version of OCRL was expressed in Sf9 cells as described (29). Native human GRB2, GRB2 dbm, and GRB3.3 with a C-terminal Myc tag were expressed as GST fusion proteins in Sf9 cells using pVIKS (25). GRB2 dbm contains mutations (P49L and E203R) in the N-and C-terminal SH3 domains that are the human counterparts of Caenorhabditis elegans Sem-5 mutations (30), and GRB3.3 contains an SH2 domain deletion (31). Mutations were generated according to Higuchi (32). Human GRB2 was expressed as a GST fusion protein in Escherichia coli using pGexKT (33).
Baculovirus Expression and Enzyme Activity-Sf9 insect cells were grown in TNM-FH medium with 10% heat-inactivated fetal calf serum and 100 g of gentamicin/ml. Approximately 3 ϫ 10 6 insect cells were infected with baculovirus encoding either a protein tyrosine phosphatase MEG-01 that served as a "negative control" (34), a 5-phosphatase, or native or mutant GRB2. For assays of enzyme activity, the cells were harvested from 60-mm dishes 3 days after infection and sonicated in 300 l of Tris, pH 7.5 (20 mM), NaCl (150 mM), MgCl 2 (3 mM), EGTA (2 mM), ␤-mercaptoethanol (10 mM), phenylmethylsulfonyl fluoride (1 mM), benzamidine (10 g/ml), pepstatin A (1 M), and leupeptin (10 g/ml). Sonicates were centrifuged at 16,000 ϫ g for 10 min and supernatants were tested for enzyme activity. SIP-110 was purified on glutathione-agarose and eluted with glutathione for use in determining K m and V max for InsP 4 hydrolysis and in heat inactivation experiments.

Proof of the Product of Hydrolysis of [ 3 H]Ins(1,3,4,5)P 4 by SIP-110 -
The products formed from incubation of SIP-110 with [ 3 H]Ins(1,3,4,5)P 4 were separated by HPLC on an Adsorbosphere SAX column equilibrated in 20 mM NH 4 H 2 PO 4 at pH 3.5. Products were eluted with a linear gradient from 0 to 0.75 M NH 4 H 2 PO 4 over 100 min. The products formed from incubation of SIP-110 product with inositol polyphosphate-1-phosphatase (40) or inositol polyphosphate-4-phosphatase (41) were separated by HPLC on a Partisil 10 SAX column equilibrated in 40 mM ammonium formate/formic acid, pH 3.5. Products were eluted with ammonium formate/formic acid, pH 3.5, 60 -400 mM for 30 min, 800 -1300 mM for 33 min, 1300 -1850 mM for the next 10 min, and 1850 -3000 for the final 46 min. One-ml fractions were collected, and radioactivity was measured in a liquid scintillation counter.
Western Blotting-All immunoblotting was done with anti-HA antibody used at 120 ng/ml. Secondary antibody was horseradish peroxidase-conjugated anti-mouse IgG from Bio-Rad (1:5000 dilution), and blots were developed using ECL (Amersham).

RESULTS
We expressed SIP-110 in Sf9 cells and assayed the soluble crude recombinant enzyme for the ability to hydrolyze soluble inositol polyphosphates. Hydrolysis of [ 3 H]Ins(1,3,4,5)P 4 to an apparent InsP 3 product was Mg 2ϩ dependent while extracts from Sf9 cells expressing an irrelevant tyrosine phosphatase as a control did not hydrolyze this substrate. We confirmed that substrate hydrolysis in Sf9 supernatants expressing SIP-110 was due to the SIP-110 by immunoprecipitating that protein with a monoclonal antibody to a hemagglutinin (HA) tag at the N terminus of the recombinant protein. Increasing amounts of antibody depleted InsP 4 hydrolyzing activity from Sf9 supernantants and resulted in appearance of activity in the protein A-Sepharose pellet (Fig. 1A). Amounts of antiserum required to immunoprecipitate InsP 4 hydrolyzing activity correlated with the amount of antiserum necessary to immunoprecipitate SIP-110 protein as determined by Western blotting with anti-HA antiserum (Fig. 1B).
We confirmed that the SIP-110 protein was responsible for both the IP 4 and PrdIns(3,4,5)P 3 hydrolyzing activity by exam-  ining the heat intactivation of purified SIP-110 at 42°C. The IP 4 hydrolyzing activity and the PtdInsP 3 hydrolyzing activity inactivate at essentially the same rate proving that the two activities reside in the same protein (Fig. 5). Further confirmation came from studies using SIP-110 with a mutation of aspartic acid 460 to alanine in the region of homology to other 5-phosphatases. A similar mutation in 5-phosphatase type II resulted in an inactive enzyme. SIP-110 D460A had no activity for either InsP 4 or PtdInsP 3 (Ref. 45, data not shown). SIP-110 was initially cloned as a GRB2-binding protein. We next determined the stoichiometry of GRB2 binding to SIP-110 and whether the binding affected enzyme activity. In the experiment shown in Fig. 6, we mixed HA-tagged SIP-110 from Sf9 cell detergent lysates with equimolar amounts of native or mutant Myc-, HA-tagged GRB2. GRB2 dbm contains an inactivating single point mutation in each SH3 domain that inhibits association of SIP-110 in COS cells (19). GRB3.3 contains an inactivating SH2 domain deletion while retaining functional SH3 domains (31). We immunoprecipitated the GRB2 with antiserum to its Myc epitope and analyzed the association of SIP-110 with GRB2 by Western blotting with anti-HA antiserum or by assaying for [ 3 H]Ins(1,3,4,5)P 4 hydrolysis. As shown in Fig. 6A, SIP-110 associates with native GRB2 in a stoichiometry of 1 GRB2:0.7 SIP-110 (0.7 Ϯ 0.1, n ϭ 5), suggesting that the stoichiometry likely is 1:1. SIP-110 failed to bind in detectable amounts to GRB2 dbm confirming that the binding of SIP-110 is via the SH3 domains of GRB2. SIP-110 also binds to GRB3.3 although less well than to native GRB2 (ratio of GRB2 to SIP-110 is 12:1). We then assayed each immunoprecipitate for [ 3 H]Ins(1,3,4,5)P 4 hydrolyzing activity. As shown in Fig. 6B with detergent lysates of GRB2 and SIP-110 and it was possible this might interfere with an effect of GRB2 on activity, we repeated them using bacterially-expressed GRB2 and SIP-110 from non-detergent cell sonicates. We assayed 0.3 pmol of SIP-110 with increasing amounts of purified bacterial native GRB2. Again there was no effect of GRB2 on hydrolysis of [ 3 H]InsP 4 by SIP-110 (data not shown). In similar experiments, we also found no effect of GRB2 on SIP-110 hydrolysis of PtdIns(3,4,5)P 3 (data not shown). Thus binding of GRB2 to SIP-110 does not affect enzyme activity implying that GRB2 serves mainly as a protein to localize SIP-110 within the cell. DISCUSSION There is a growing family of inositol polyphosphate-5-phosphatases that hydrolyze one or more of the 5-phosphate containing soluble and lipid inositol phosphates. They can be grouped according to their substrate specificity. Group I 5-phosphatases have molecular masses of 32-43 kDa and hydrolyze both Ins(1,4,5)P 3 and Ins(1,3,4,5)P 4 but neither PtdIns(4,5)P 2 nor PtdIns(3,4,5)P 3 . Originally purified from platelet cytosol as type I 5-phosphatase (35,46), cDNAs encoding enzymes with similar characteristics have been cloned from a variety of tissue sources (26,27,47). Immunoprecipitation of 5-phosphatase activity from human platelets with antisera to the bovine brain and human placental isozymes suggest that these type I activities may be the same enzyme (48,49).
Group IV 5-phosphatases currently have one member: the 110-kDa SIP-110 (20) characterized in this article and the alternatively spliced 133-kDa SHIP or SIP-130 (20 -23). This 5-phosphatase hydrolyzes only 3-phosphate-containing inositol phosphates, Ins(1,3,4,5)P 4 and PtdIns(3,4,5)P 3 . In addition to the 5-phosphatases described above, there are at least three additional family members based on homologous amino acid sequence derived from cDNA clones: INPPL1 that is closely related to SIP-110 (55) and at least two expressed sequence tags contributed to GenBank.
One of the unusual characteristics of SIP-110 is its ability to associate with GRB2. Experiments reported here show that the binding of SIP-110 to GRB2 has a stoichiometry of approximately 1:1 and confirm that mutations in the SH3 domains of GRB2 can eliminate binding to SIP-110. Another protein related to phosphatidylinositol metabolism also binds to GRB2. The p85 subunit of PtdIns 3-kinase (56) was found to be associated with GRB2 SH3 domains independently of growth factor stimulation. Sos, a guanine nucleotide exchange factor for Ras, also associates with the GRB2 SH3 domains (57)(58)(59). GRB2 is proposed to function by bringing its associated proteins into complexes with other tyrosine-phosphorylated proteins such as tyrosine-phosphorylated receptors or tyrosine-phosphorylated Shc (59 -61). Studies with GRB2 and Sos suggest that binding of GRB2 to Sos does not affect the guanine nucleotide exchange activity of Sos, and that localizing Sos to the cell membrane by other mechanisms allows for full activation of Ras by Sos (61)(62)(63). Our results with SIP-110 suggest a similar role for GRB2. We find that GRB2 binding has no effect on SIP-110 activity in vitro, suggesting that GRB2 serves to localize SIP-110 into complexes with other proteins and/or to allow for SIP-110 to associate with the cell membrane. Since one of the two SIP-110 substrates is the inositol lipid, PtdIns(3,4,5)P 3 , GRB2 association with a membrane receptor would allow for greater access of SIP-110 to this substrate. While other 5-phosphatases have a higher first-order rate constant for hydrolysis of PtdIns(3,4,5)P 3 in vitro, localization of SIP-110 in areas of PtdIns(3,4,5)P 3 concentration may allow for efficient hydrolysis of this substrate.
Since GRB2 and two other constitutively associated proteins, Sos and the PI 3-kinase, are implicated in regulation of the Ras pathway (39, 59 -61, 64), and both SIP-110 substrates have also been linked to Ras activation (3,61), we speculate that SIP-110 may have a role in regulating Ras activity. One substrate, Ins(1,3,4,5)P 4 , has been shown to bind a GAP1-like protein and stimulate its Ras GAP activity under some conditions in vitro (3). Hydrolysis of Ins(1,3,4,5)P 4 by SIP-110 would decrease this Ras GAP1 activity, resulting in a prolonged activation of Ras. Likewise, a number of studies have suggested that activation of the PI 3-kinase either precedes or follows Ras activation (39,64). Since SIP-110 hydrolyzes one product of an active PI 3-kinase, PtdIns(3,4,5)P 3 , hydrolysis of this substrate would be predicted to affect the activity of the Ras pathway. Activation of the PI 3-kinase has been shown activate c-Akt or protein kinase B (18,19), but there are also several studies showing that members of the protein kinase C family of enzymes are activated by inositol containing phospholipids (65)(66)(67). In situations in which PtdIns(3,4,5)P 3 has a positive effect, its hydrolysis by SIP-110 would have negative regulatory consequences. However, recent studies on the phosphorylation of pleckstrin in platelets show that the major phase of thrombin-stimulated phosphorylation is inhibited by the PtdIns 3-kinase inhibitor, wortmannin (68,69). This phosphorylation may correlate with the production of PtdIns(3,4)P 2 rather than PtdIns(3,4,5)P 3 , and addition of PtdIns(3,4)P 2 in saponin-permeabilized platelets can mimic the effect of thrombin in stimulating pleckstrin phosphorylation (66). Likewise, in vitro studies with Akt have demonstrated that PtdIns(3,4)P 2 activates Akt by binding to its pleckstrin homology domain (69). In situations such as this where PtdIns(3,4)P 2 is an activating signal, SIP-110 hydrolysis of PtdIns(3,4,5)P 3 would likewise have an activating effect.
In this study we examine the enzyme activity and products of recombinant 110-kDa SIP-110 and the effects of GRB2 on its activity. This is the newest member of a multigene enzyme family that is likely to influence multiple cell signaling pathways.