Regulation of phosphatidylinositol 3,4,5-trisphosphate 5'-phosphatase activity by insulin.

Polyphosphoinositides are thought to be mediators of cellular signaling pathways as well as regulators of cytoskeletal elements and membrane trafficking events. It has recently been demonstrated that a class of phosphatidylinositol (PI) 3,4,5-P3 5′-phosphatases contains SH2 domains and proline-rich regions, which are present in many signaling proteins. We report here that insulin stimulation of Chinese hamster ovary cells (CHO-T) expressing human insulin receptors causes an 8-10-fold increase in PI 3,4,5-P3 5′-phosphatase activity in anti-phosphotyrosine immunoprecipitates of the cell lysates. This insulin-sensitive polyphosphoinositide 5′-phosphatase did not catalyze dephosphorylation of PI 4,5-P2. No change in 5′-phosphatase activity was detected in insulin receptor or IRS-1 immune complexes in response to insulin. However, insulin treatment of CHO-T cells markedly increased the PI 3,4,5-P3 5′-phosphatase activity associated with Shc and Grb2. The insulin-regulated polyphosphoinositide 5′-phosphatase was not immunoreactive with antibody raised against the recently cloned SHIP 5′-phosphatase reported to associate with Shc and Grb2 in B lymphocytes. These data demonstrate that insulin causes formation of complexes containing a PI 3,4,5-P3 5′-phosphatase, and Shc or Grb2, or both, suggesting an important role of this enzyme in insulin signaling.

The insulin receptor belongs to a family of structurally related transmembrane growth factor receptors that exhibit ligand-activated protein-tyrosine kinase activity (1)(2)(3). The insulin receptor kinase activity is thought to be essential for cellular responses to insulin (4 -6). Activation of insulin receptor kinase promotes the rapid autophosphorylation of insulin receptor ␤-subunits as well as tyrosine phosphorylation of several cytoplasmic proteins such as IRS-1, 1 Shc, and pp60 that appear to be involved in the insulin signaling pathway (3,(7)(8)(9). Evidence indicates that a primary function of the insulin receptor kinase is to place tyrosine phosphate docking sites on these proteins for the recruitment of signaling proteins containing Src homology (SH) 2 domains (1,10,11). Thus, insulininduced phosphorylation of IRS-1, Shc, and pp60 promotes their association with specific SH2-containing proteins, which in turn can stimulate the catalytic activity of these SH2 proteins (12)(13)(14)(15). One such SH2-containing protein is the p85 regulatory subunit of the p110 phosphatidylinositol (PI) 3-kinase which catalyzes phosphorylation of the 3-position on PI (12,13,16).
Strong evidence supports a pivotal role for signaling complexes containing IRS-1 and the p85/p110-type PI 3-kinases in mediating insulin action on GLUT4 glucose transporter redistribution to the plasma membrane leading to increased glucose uptake as well as glycogen synthesis. Inhibition of PI 3-kinase activity by wortmannin (17)(18)(19) or LY294002 (20), microinjection of a fusion protein consisting of an SH2 domain of the p85 regulatory subunit of PI 3-kinase (21), and disruption of PI 3-kinase recruitment to IRS-1 by dominant inhibitory constructs of p85 (22) ablate the stimulation of glucose transport and glycogen synthesis by insulin. Expression of IRS-1 antisense RNA in isolated fat cells also inhibits insulin-mediated translocation of epitope-tagged GLUT4 glucose transporters to the cell surface (23). Further, insulin causes the localization of IRS-1⅐PI 3-kinase complexes to intracellular membrane vesicles containing GLUT4 (24), while other growth factors that stimulate PI 3-kinase activity but fail to activate glucose transport do not. 2 These data are consistent with the hypothesis that one or more 3Ј-phosphoinositide species generated in intracellular membranes in response to insulin regulate cellular components involved in membrane trafficking of GLUT4.

EXPERIMENTAL PROCEDURES
Material-4G10 anti-phosphotyrosine (anti-Tyr(P)) mouse monoclonal and anti-p85 polyclonal antibodies were purchased from UBI and used according to manufacturer's specifications. Rabbit polyclonal anti-IRS-1 immunoglobulin used for immunoprecipitation was prepared by injecting a peptide of the COOH-terminal 15 amino acids derived from the sequence of rat IRS-1 conjugated to keyhole limpet hemocyanin into New Zealand White rabbits. An IgG fraction from the resultant serum was prepared by Protein A-Sepharose chromatography. Insulin receptor-specific monoclonal antibody CT-1 was from mouse ascites (a kind gift of Dr. Ken Siddle). Anti-Shc monoclonal and polyclonal antibodies were purchased from Transduction Laboratories. Anti-Grb2 polyclonal antibody was from Santa Cruz. Phosphatidylinositol and phosphatidylserine were from Avanti Polar Lipids. PI 4-P and PI 4,5-P 2  Cell Culture-CHO-T cells were maintained in Ham's F-12 media, 10% fetal bovine serum, 50 g/ml streptomycin/penicillin, and grown to confluence before use.
Cell Lysis and Immunoprecipitation-CHO-T cells were serumstarved for 18 -24 h, stimulated with 100 nM insulin for 15 min, and then washed twice on ice in phosphate-buffered saline (PBS). Cells were then lysed in ice-cold lysis buffer composed of 0.5% Nonidet P-40, 50 mM Hepes pH 7.4, 100 mM NaF, 10 mM NaPP i , 2 mM Na 3 VO 4 , 1 mM phenylmethylsufonyl fluoride, 40 g/ml aprotinin, 40 g/ml leupeptin. After 20 min of incubation on ice, insoluble material was removed by centrifugation (14,000 ϫ g for 15 min). The supernatant was removed and assayed for total protein content, using a BCA protein determination kit with bovine serum albumin as standard. Equal amounts of protein (typically 1 mg of protein) from each cleared lysate were then incubated overnight at 4°C on a end-over-end mixer with 4 g of anti-Tyr(P), 60 g of anti-IRS-1 IgG, 25 g of anti IR, 4 g of anti-Shc, 1 g of anti-Grb2, or 2 l of anti-SHIP antibodies. Mouse or rabbit antibodies were then adsorbed to anti-mouse IgG-Sepharose (40 l) or Protein A-Sepharose (40 l), respectively, for 2 h at 4°C by end-overend mixing. The Sepharose-bound immune complexes were then collected by centrifugation (10,000 ϫ g, 30 s) and washed five times in ice-cold PBS containing 0.1% Nonidet P-40. The samples were then used in immunoblot experiments or in PI 3,4,5-P 3 phosphatase assays.

RESULTS AND DISCUSSION
Initial experiments were conducted to establish assay conditions for quantifying polyphosphoinositide 5Ј-phosphatase activity in lysates of CHO-T cells expressing human insulin receptors. Dephosphorylation of [3Ј-32 P]PI 3,4,5-P 3 to di-and monophosphoinositides in cell lysates was readily observed in the absence of vanadate upon thin layer chromatography analysis of the reaction products (not shown). However, inclusion of vanadate in the assay buffer almost completely blocked PI 3,4-P 2 phosphatase activity with little effect on the PI 3,4,5-P 3 5Ј-phosphatase activity. This is consistent with a previous report that 3Ј-phosphatase is inhibited by vanadate (27). Thus, in the presence of vanadate, nearly quantitative conversion of PI 3,4,5-P 3 to PI 3,4-P 2 was observed, measured either as the disappearance of the former or appearance of the latter (not shown). All subsequent assays were therefore performed under these conditions.
To determine whether insulin regulates PI 3,4,5-P 3 5Ј-phosphatase, lysates of CHO-T cells incubated with or without insulin were immunoprecipitated with anti-Tyr(P) antibody, and the precipitates were assayed as described above. The results revealed a marked increase in PI 3,4,5-P 3 5Ј-phosphatase activity in the anti-Tyr(P) precipitates due to insulin action (Fig. 1, A and B). Tyrosine-phosphorylated insulin receptor and IRS-1 were immunoprecipitated when cells were treated with insulin under these conditions, as evidenced by Western blot analysis (Fig. 1C). However, when either insulin receptors or IRS-1 were specifically immunoprecipitated with anti-insulin receptor or anti-IRS-1 antibody (Fig. 1C), no insulin-stimu-

FIG. 1. Insulin regulates PI 3,4,5-P 3 5-phosphatase.
Lysates were prepared from CHO-T cells treated with (ϩ) or without (Ϫ) 100 nM insulin for 15 min. A, cell lysates were immunoprecipitated with antiphosphotyrosine (Anti-p-tyr), anti-insulin receptor (Anti-IR), or anti-IRS-1 antibodies and the immune complexes were subjected to a PI 3,4,5-P 3 5Ј-phosphatase assay by incubating with [ 32 P]PI 3,4,5-P 3 , and the reaction products were analyzed by thin layer chromatography. NA indicates parallel tubes containing PI 3,4,5-P 3 , but with no addition of immune complexes. B, the graph shows the effect of insulin on PI 3,4,5-P 3 5Ј-phosphatase activity in immunoprecipitates. Spots corresponding to PI 3,4-P 2 formed upon thin layer plate were cut out and quantitated using a ␤-counter. The data presented are the average values from 3 independent experiments Ϯ S.E. C, cell lysates were immunoprecipitated with anti-phosphotyrosine, anti-IR, or anti-IRS-1, and immune complexes were resolved by SDS-PAGE on a 7.0% gel and electrophoretically transferred to nitrocellulose for 12 h at 125 mA. The filter was blocked and incubated with anti-Tyr(P) antibody and then horseradish peroxidase-anti-mouse followed by detection with chemiluminescence.
lated PI 3,4,5-P 3 5Ј-phosphatase activity could be detected in the immune complexes (Fig. 1, A and B). This was also the case when whole cell lysates from control versus insulin-treated CHO-T cells were assayed (data not shown).
The insulin-regulated PI 3,4,5-P 3 5Ј-phosphatase activity present in the anti-tyrosine phosphate immunoprecipitates was dependent on MgCl 2 in the assay buffer (data not shown). The anti-Tyr(P) immunoprecipitates did not contain insulinregulated phosphatase activity when PI 4,5-P 2 was used as substrate (data not shown). High pressure liquid chromatography of the deacylated 32 P-labeled phosphoinositide reaction products was performed to identify the diphosphoinositide formed in the presence of tyrosine phosphate immune complexes from control and insulin-treated CHO-T cells (Fig. 2). This analysis demonstrated virtually quantitative conversion of PI 3,4,5-P 3 to PI 3,4-P 2 in this assay as well as a marked increase in PI 3,4-P 2 formation catalyzed by the immune complexes derived from insulin-treated cells (Fig. 2).
The recent molecular cloning of polyphosphoinositide 5Јphosphatases that associate with Shc in response to myeloid cell activation (30,31) and that associate with the adapter protein Grb2 in response to B cell activation, prompted experiments to test such possible associations with the insulin-regulated 5Ј-phosphatase(s) (Fig. 3). Immune complexes obtained from lysates of insulin-treated CHO-T cells using anti-Shc or anti-Grb2 antibodies exhibited PI 3,4,5-P 3 5Ј-phosphatase activity that was severalfold higher than that derived from control cells. However, the insulin-stimulated 5Ј-phosphatase activity in the anti-Shc and anti-Grb2 immunoprecipitates was about half that associated with the anti-Tyr(P) antibody (Fig.  3). This probably reflects, at least in part, the fact that the anti-Shc and anti-Grb2 antibodies used do not quantitatively precipitate Shc and Grb2 from the cell lysates (data not shown). Taken together, the data in Figs. 1-3 demonstrate a marked effect of insulin to cause association of Shc and Grb2 with one or more 5Ј-phosphatases able to dephosphorylate PI 3,4,5-P 3 but not PI 4,5-P 2 .
The characteristics of this insulin-regulated polyphosphoinositide 5Ј-phosphatase resembles those of SHIP, the recently cloned 5Ј-phosphatase that binds to the C-terminal SH3 domain of Grb2 in vitro and associates with Shc in response to cytokines in hemopoietic cells (30). This enzyme also requires Mg 2ϩ for activity. The availability of anti-SHIP antibody (30) allowed us to test whether SHIP was the insulin-regulated 5Ј-phosphatase detected in the present experiments. CHO-T cell lysates were immunoprecipitated with anti-SHIP antibody, and both the precipitate and supernatant were assayed for PI 3,4,5-P 3 5Ј-phosphatase activity. As shown in Fig. 4, only the latter catalyzed formation of labeled PI 3,4-P 2 in this assay. Parallel immunoprecipitations of hamster lung lysates with anti-SHIP revealed easily detectable PI 3,4,5-P 3 5Ј-phosphatase in the precipitates as well as in the supernatants (Fig. 4). Similarly, the anti-SHIP precipitates from hamster lung displayed a 150-kDa band upon SDS-PAGE and immunoblotting that was not present in anti-Tyr(P) imunoprecipitates of insulin-treated CHO-T cells lysates (data not shown). These latter immune complexes exhibited significant PI 3,4,5-P 3 5Ј-phosphatase activity as demonstrated in Figs. 1-3. Thus, these data indicate that the insulin-regulated polyphosphoinositide 5Јphosphatase is not SHIP itself. However, it is possible that the CHO-T cell 5Ј-phosphatase is an isoform of this enzyme that is not recognized by anti-SHIP antibody. Resolving the identity of the presently described insulin-regulated PI 3,4,5-P 3 5Ј-phosphatase is an important future objective.
It is significant that the polyphosphoinositide 5Ј-phosphatases described above (30,31) and reported here require phosphorylation of the 3Ј-position of the phosphoinositol head group to catalyze dephosphorylation. Phosphorylation of this 3Ј-position is catalyzed by four known classes of PI 3-kinases that exhibit diverse regulatory mechanisms (16,36,38,39). Thus, the regulated phosphorylation of PI 4,5-P 2 by PI 3-kinases theoretically provides substrate for the 5Ј-phosphatases specific for PI 3,4,5-P 3 . These two reactions catalyzed by PI 3-ki- A, cell lysates were immunoprecipitated with anti-Tyr(P), anti-Shc, or anti-Grb2, and immunoprecipitates were assayed for PI 3,4,5-P 3 5Јphosphatase by incubation with [ 32 P]PI 3,4,5-P 3 . NA indicates assay, without immune complex. B, graph shows the effect of insulin on PI 3,4,5-P 3 5Ј-phosphatase activity in immune complexes. Spots corresponding to PI 3,4-P 2 in the thin layer chromatography plate were cut out and quantitated using a ␤-counter. The data presented are average values from 3 independent experiments Ϯ S.E., normalized by equating each value in the absence of insulin to 1. The actual counts/min values in basal conditions were: 1014, anti-Tyr(P); 313, anti-Shc; and 890, anti-Grb2. nase and PI 3,4,5-P 3 5Ј-phosphatase in combination promote the conversion of PI 4,5-P 2 to PI 3,4-P 2 . The fact that insulin and other growth factor receptor tyrosine kinases regulate both of these enzymes indicates that at least part of the cellular PI 3,4-P 2 generated in response to these signaling pathways derives from newly formed PI 3,4,5-P 3 . Further, these considerations strongly suggest an important role for PI 3,4-P 2 in signaling by these receptors. PI 3,4-P 2 may be an effector that is unique and specific for one or more downstream signaling pathways such as the cAkt/Rac protein kinase (40). Thus, the PI 3,4,5-P 3 5Ј-phosphatase reaction may serve as a branch point for polyphosphoinositide signaling in which PI 3,4,5-P 3 activates a set of events distinct from those activated by PI 3,4-P 2 . It will be important to search for cellular proteins that specifically bind each of these phosphoinositides.
The association of 5Ј-phosphatase with Shc and Grb2 in response to insulin reported here (Fig. 3) suggests a potential role of this phosphatase in regulating the p21 ras pathway. Both Shc and Grb2 are components of multiprotein complexes containing the guanine nucleotide exchange factor son of sevenless that catalyzes GTP loading of p21 ras (8,11,41,42). Interactions between p21 ras and the p110 PI 3-kinase have been reported (43), and in some cell types a downstream effect of p21 ras , mitogen-activated protein kinase activation, in response to insulin or growth factors is blocked by wortmannin, a potent inhibitor of PI 3-kinases (44). Futher studies on this issue are clearly warranted.
In addition to a potential positive signaling function of the insulin-sensitive PI 3,4,5-P 3 5Ј-phosphatase described above, a negative role in signal transmission is also possible. If PI 3,4,5-P 3 generated by PI 3-kinases is a positive effector of downstream signaling events, as appears likely (16), its concentration is expected to be reduced by the action of the 5Јphosphatase. Thus, the signaling potential of PI 3,4,5-P 3 may be reduced or desensitized by insulin-regulated PI 3,4,5-P 3 5Ј-phosphatase. Perhaps it is desirable to control the cellular localization of PI 3,4,5-P 3 and to restrict it from regions containing Shc⅐Grb2 complexes. This hypothesis requires rigorous testing subsequent to identification of additional cellular targets of PI 3,4,5-P 3 . In any case, the data reported here indicate an important function of PI 3,4,5-P 3 5Ј-phosphatase activity in one or more signaling pathways emanating from the insulin receptor.