Type I Transforming Growth Factor β Receptor Binds to and Activates Phosphatidylinositol 3-Kinase*

We have examined the interaction of transforming growth factor (TGF)β receptors with phosphatidylinositol 3-(PI3) kinase in epithelial cells. In COS7 cells, treatment with TGFβ increased PI3 kinase activity as measured by the ability of p85-associated immune complexes to phosphorylate inositides in vitro. Both type I and type II TGFβ receptors (TβR) associated with p85, but the association of TβRII appeared to be constitutive. The interaction of TβRI with p85 was induced by treatment with TGFβ. The receptor association with PI3 kinase was not direct as 35S-labeled rabbit reticulocyte p85 did not couple with fusion proteins containing type I and type II receptors. A kinase-dead, dominant-negative mutant of TβRII blocked ligand-induced p85-TβRI association and PI3 kinase activity. In TβRI-null R1B cells, TGFβ did not stimulate PI3 kinase activity. This stimulation was restored upon reconstitution of TβRI by transfection. In R1B and NMuMG epithelial cells, overexpression of a dominant active mutant form of TβRI markedly enhanced ligand-independent PI3 kinase activity, which was blocked by the addition of the TβRI kinase inhibitor LY580276, suggesting a causal link between TβRI function and PI3 kinase. Overexpressed Smad7 also prevented ligand-induced PI3 kinase activity. Taken together, these data suggest that 1) TGFβ receptors can indirectly associate with p85, 2) both receptors are required for ligand-induced PI3 kinase activation, and 3) the activated TβRI serine-threonine kinase can potently induce PI3 kinase activity.

Transforming growth factor ␤ (TGF␤) 1 binds to a heteromeric complex of transmembrane serine-threonine kinases, the type I and II TGF␤ receptors (T␤RI or Alk5 and T␤RII). Following ligand binding to T␤RII, the type I receptor is recruited to the ligand-receptor complex where the constitutively active T␤RII transactivates T␤RI (1). Activated T␤RI phosphorylates the receptor-specific Smad2 and Smad3, which then associate with Smad4 and, as a heteromeric complex, translocate to the nucleus where they regulate the transcription of TGF␤ target genes (1). The Smad signaling pathway mediates the antiproliferative effect of TGF␤ in epithelial cells (2,3) and is the best characterized. Several non-Smad pathways have also been implicated in mediating the cellular effects of TGF␤. These include the extracellular signal-regulated kinase (Erk), c-Jun N-terminal kinase (Jnk), p38 mitogen-activated protein kinase (MAPK), phosphatidylinositol-3 (PI3) kinase, and the family of Rho GTPases (4,5).
Activated PI3 kinase increases the formation of intracellular 3Ј-phosphorylated inositol lipids, signal transducers that are essential for the regulation of cell cycle progression, glucose metabolism, cell motility, epithelial-to-mesenchymal transition, and apoptosis among others (6,7). In Swiss 3T3 cells, TGF␤ stimulates PI3 kinase activity, as measured by the ability of immune complexes precipitated with an antibody against p85, the regulatory subunit of PI3 kinase, to induce the formation of phosphatidylinositol-3 monophosphate (PI3P) in vitro (8). In mammary epithelial cells, TGF␤ induces epithelial-tomesenchymal transition and cell motility as well as phosphorylation and activation of the PI3 kinase-dependent Akt serinethreonine kinase. These responses are blocked by LY294002, a small molecule inhibitor of the p110 catalytic subunit of PI3 kinase (9). In mesenchymal and epithelial cells, TGF␤-mediated protection from apoptosis is blocked by LY294002 and/or expression of dominant-negative, kinase-dead Akt (10 -12). Inhibition of PI3 kinase reverses the fibroblastoid phenotype of TGF␤-treated Ras-transformed hepatocytes to an epithelial phenotype (13). Furthermore, PI3 kinase inhibition abrogates both basal and TGF␤-induced motility in mammary cancer cells (9). Taken together, these data suggest that PI3 kinase is a major effector pathway of the transforming effects of TGF␤ in epithelial cells. Therefore, in this study, we have examined the initial mechanisms of TGF␤-induced activation of PI3 kinase in epithelial cells.

MATERIALS AND METHODS
Cell Lines and Antibodies-COS7 and NMuMG cells were purchased from American Type Culture Collection (ATCC). R1B cells were provided by Harold Moses (Vanderbilt University, Nashville, TN). COS7 and R1B cells were maintained in Dulbecco's Modified Essential Medium (Invitrogen) supplemented with 10% fetal bovine serum (Hyclone). NMuMG cells were maintained in 10% fetal bovine serum-Dulbecco's modified Eagle's medium with 10 g/ml insulin. We utilized the following antibodies: p85 (Upstate Biotechnology), Smad2/3 (Transduction Laboratories), hemagglutinin (HA) rabbit polyclonal and T␤RI (Santa Cruz Biotechnology), FLAG epitope (Sigma), Ser-473 P-Akt, total Akt, P-MAPK, total MAPK (Erk42/44), and P-Smad2 (Cell Signaling). Human recombinant TGF␤1 was obtained from R&D Systems. The PI3 kinase inhibitor wortmannin was from Calbiochem. The T␤RI kinase inhibitor LY580276 was provided by Jonathan Yingling (Lilly * This work was supported by National Institutes of Health Grant R01 CA62212 (to C. L. A.), Breast Cancer Specialized Program of Research Excellence (SPORE) Grant P50 CA98131, and Vanderbilt-Ingram Comprehensive Cancer Center Support Grant CA68485. 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.

Research Laboratories, Indianapolis, IN).
Plasmids and Viral Vectors-The pCMV6-p85 construct tagged with c-myc in its C terminus was provided by Lewis Cantley (Harvard Medical School, Boston, MA). The pRK5-T␤RII-Flag plasmid was provided by Rik Derynck. The retroviral vector pGabe-T␤RII K277R -HA was from Martin Oft (both at the University of California at San Francisco) and has been described previously (14). The wild-type T␤RI-HA construct, from Kohei Miyazono (Japanese Foundation for Cancer Research, Tokyo), was subcloned into the retroviral expression vector pBMN-IRES-EGFP (provided by Garry Nolan, Stanford University, Palo Alto, CA) (15). Cells were transfected with plasmids using Fu-GENE 6 reagent (Roche Applied Sciences) according to the manufacturer's instructions. Retroviruses were prepared by transfection of 293T cells with 15 g of DNA/100-mm dish of three plasmids encoding gag/ pol, VSV-G, and the target construct (in 4:3:8 ratios, respectively). Supernatants from cells were collected for 2 days and combined, filtered through 0.4-m filters, and stored in aliquots at Ϫ80°C. R1B cells were infected with supernatant containing retroviruses in the presence of 8 g/ml Polybrene (Sigma) as described previously (16). Three days after transduction of R1B cells with pBMN-IRES-EGFP, EGFP-positive cells were selected by flow cytometry. Under these conditions, Ͼ95% of selected cells expressed GFP at the time of any of our experiments. The HA-Alk5 T204D and Flag-Smad7 adenoviral constructs were also from Dr. Kohei Miyazono. Stocks of recombinant adenoviruses were generated in 293T cells and titered utilizing the Takara assay (Takara Biomedicals, Tokyo, Japan). Cells were then infected with these or with a control ␤-galactosidase adenovirus at an equivalent multiplicity of infection as described previously (17).
Immunoprecipitation and Immunoblot Analysis-Cells were washed twice with phosphate-buffered saline and lysed in 1% Nonidet P-40 buffer (1% Nonidet P-40, 120 mM NaCl, and 50 mM Tris-HCl, pH 7.4, protease inhibitor mixture (Roche Applied Science), 20 mM NaF, 1 mM Na 3 VO 4 ) on ice for 20 min. Cell lysates were clarified by centrifugation at 14,000 rpm at 4°C for 20 min, and protein concentrations were then determined by the BCA method (Pierce). Equal amounts of cell lysates were incubated overnight at 4°C with primary antibodies. Immune complexes were collected with protein G-Sepharose 4B or protein A-Sepharose 4B (both from Sigma) at 4°C for 3 h. The precipitates were pelleted by centrifugation and washed three times with lysis buffer, once with high salt buffer (10 mM Tris-HCl, pH 7.4, 0.5 M NaCl) and once with low salt buffer (10 mM Tris-HCl, pH 7.4). The final pellets were boiled in Laemmli buffer containing 10% ␤-mercaptoethanol and separated by SDS-PAGE.
For immunoblot analysis, 20 -30 g of total cell lysates were separated by SDS-PAGE and transferred to nitrocellulose membranes (Bio-Rad). Membranes were blocked with 5% nonfat dry milk in TBST (Tris-buffered saline containing 0.1% Tween 20) at room temperature for 1 h. The blots were incubated with primary antibody in 4°C for 16 h, washed three times with TBST for 1 h, incubated for 2 h with secondary anti-mouse or anti-rabbit antibody conjugated with horseradish peroxidase (Amersham Biosciences), and then washed three times with TBST for an additional 1 h. Immunoreactive bands were visualized using an enhanced chemiluminescence (ECL) detection system (Amersham Biosciences) as described (17).
GST Pull-down Assays-The expression vectors for glutathione Stransferase (GST) fusion proteins, pGEX4T and pGEX4T-T␤RI (a gift from Pran Datta, Vanderbilt University) and pGEX2T-T␤RII (provided by Rik Derynck, UCSF) were transformed into Escherichia coli BL21 strain. The bacterial cultures were induced by 0.1 mM isopropyl-Dthiogalactopyranoside for 5 h at 37°C. E. coli were pelleted, washed with STE buffer (10 mM Tris, pH 8.0, 150 mM NaCl, 1 mM EDTA), and incubated in STE buffer containing 100 g of lysozyme for 15 min on ice. Protease inhibitors, 5 mM dithiothreitol and 1.5% sarcosyl were added to the resuspended E. coli, which were next sonicated. Bacterial lysates were brought up in 4% Triton, 10 g of DNase, and 8 mM MgCl 2 and incubated for 2 h at 4°C followed by centrifugation at 12,000 ϫ g for 5 min. GST fusion proteins were purified by binding to glutathione (GSH)-Sepharose 4B beads (Amersham Biosciences) at room temperature for 1 h before washing three times with phosphate-buffered saline containing protease inhibitors. The concentration of proteins immobilized on beads was estimated by SDS-PAGE and semi-quantitated against bovine serum albumin standards (Sigma) after staining the SDS gel with Coomassie Blue. Binding assays were performed between GST fusion proteins and in vitro translated p85 protein. p85 was generated with a TNT-coupled reticulocyte lysate system (Promega) also containing pCMV6-p85-myc, T7 polymerase, and 35 S-labeled methionine (Amersham Biosciences). 10 l of 35 S-labeled p85 protein and equal amounts of immobilized GST or GST fusion proteins were incu-bated for 2 h at 4°C with gentle rotation. The precipitates were pelleted by centrifugation and washed three times with 1% Nonidet P-40 buffer, once with high salt buffer and once with low salt buffer. The beads were resuspended in Laemmli buffer, boiled, separated by SDS-PAGE, and analyzed by autoradiography.
PI3 Kinase Assay-PI3 kinase catalytic activity was measured in vitro as described previously (18). In brief, cells were lysed with 1% Nonidet P-40 in buffer A (137 mM NaCl, 20 mM Tris-HCl, pH 7.4, 1 mM CaCl 2 , 1 mM MgCl 2 , and 0.1 mM Na 3 VO 4 ) and a protease inhibitor mixture. Lysates were precipitated with a p85 antibody and protein A-Sepharose 4B. After 2 washes in 1% Nonidet P-40 in buffer A and 2 washes in assay buffer (20 mM Hepes, 150 mM NaCl, 0.1 mM EDTA), the precipitates were resuspended in 80 l of assay buffer containing 10 l of crude brain phosphoinositides (PI, 2 mg/ml; Sigma) and 10 l of substrate (500 M cold ATP with 10 Ci/1 l of [␥-32 P]ATP (specific activity 6000 Ci/mmol; Amersham Biosciences) in 200 mM Hepes, pH 7.5, and 50 mM MgCl 2 ). Following gentle agitation for 10 min at room temperature, the reaction was terminated by the addition of 100 l of 1 N HCl and 200 l of chloroform:methanol (1:1). The radiolabeled lipids were extracted, concentrated, and separated by TLC using silica gel plates pretreated with 1% w/v K oxalate in a solvent system of npropanol:2 M acetic acid (65:35 v/v). The incorporation of 32 P into PI was detected by autoradiography.

RESULTS
p85, the Regulatory Subunit of PI3 Kinase, Can Bind Both T␤RII and T␤RI-In serum-starved COS7 cells, addition of TGF␤ stimulated PI3 kinase activity in a time-dependent fashion as measured by the ability of immune complexes precipitated with p85 antibodies to stimulate formation of 3Ј-phosphorylated inositol lipids in vitro. This induction was maximal at 6 h and temporally correlated with phosphorylation of the serine-threonine kinase Akt in Ser-473 (Fig. 1). The kinetics of ligand-induced Smad2 phosphorylation were faster with maximal induction seen after treatment with TGF␤ for 1 h.
To determine whether p85 interacted directly with TGF␤ receptors, we co-transfected COS7 cells with vectors encoding p85 and T␤RI or T␤RII followed by immunoprecipitation with FIG. 1. TGF␤ stimulates PI3 kinase activity in COS7 cells. COS7 cells were serum-starved for 24 h and then treated with 2 ng/ml TGF␤ for the indicated times. Left panel, cells were lysed as described under "Materials and Methods." Cell lysates were next precipitated with an antibody against p85 and protein A-Sepharose beads. The immune complexes were tested in a PI3 kinase in vitro assay as described under "Materials and Methods." Where indicated, wortmannin (Wn, 500 nM) was added to the in vitro reaction as a negative control. Right panel, total cell lysates (without immunoprecipitation) were separated by SDS-PAGE and tested by immunoblot using the antibodies indicated to the right of the panel. receptor antibodies. p85 was detectable in T␤RI and T␤RII pull downs from cells co-transfected with the respective combination of vectors ( Fig. 2A). To determine whether this binding is modulated by TGF␤, COS7 cells were stimulated with TGF␤ for variable times. The binding between p85 and T␤RI was maximal 1 h after TGF␤ treatment, and this association was correlated with maximal induction of Smad2 and Akt phosphorylation (Fig. 2B, left). However, binding between p85 and T␤RII was not affected by exogenous TGF␤ even though in these co-transfected cells a more modest ligand-induced phosphorylation of Smad2 and Akt was still observed (Fig. 2B,  right). To determine whether this p85-TGF␤ receptor association was direct, we co-incubated rabbit reticulocyte 35 S-labeled p85 with GST-T␤RI and with GST-T␤RII fusion proteins (Fig.  2C). Under these conditions, we were unable to recover labeled p85 from the immobilized GST-TGF␤ receptor fusions, implying that the p85-T␤RI and p85-T␤RII associations, as supported by the prior co-precipitation experiments in transfected cells (Fig. 2, A and B), were not direct.
T␤RII and T␤RI Are Required for Ligand-mediated Activation of PI3 Kinase-To investigate whether T␤RII function is required for ligand-induced activation of PI3 kinase, we cotransfected COS7 cells with p85 and T␤RI with or without a vector encoding a kinase-inactive T␤RII mutant in which the lysine at position 277 has been mutated to arginine (14). The expression of the T␤RII K277R construct prevented ligand-induced association of p85 with T␤RI (Fig. 3A). In cells co-transfected with p85 and T␤RI, we were unable to detect an increase in TGF␤-induced PI3 kinase activity associated with p85 precipitates (data not shown). We speculate that under these con-ditions, the majority of the overexpressed p85 protein is not associated with T␤RI, and thus, the p85-associated catalytic activity does not reflect ligand-induced, TGF␤ receptor-associated PI3 kinase activity. Therefore, we transfected T␤RI without p85 but with or without T␤RII K277R . The expression of the kinase-dead type II receptor abrogated a ligand-mediated increase in PI3 kinase activity in vitro (Fig. 3B). This is different from the slower kinetics of ligand-induced activation observed in untransfected COS7 cells (Fig. 1). In COS7 cells transfected with T␤RI, but not with T␤RII, exogenous TGF␤ maximally increased PI3 kinase activity at 1 h. We did not detect any PI3 kinase activity in T␤RI precipitates.
To examine whether T␤RI was required for ligand-induced activation of PI3 kinase, we used R1B cells, a subline of Mv1Lu cells, which lacks type I receptors (19). R1B cells were transduced with pBMN-IRES-EGFP or pBMN-T␤RI (Alk5)-IRES-EGFP retroviral vectors; EGFP-positive cells were sorted by flow cytometry. In all our experiments, Ͼ95% of cells used expressed GFP. In T␤RI-reconstituted cells but not in cells transduced with the control retrovirus, treatment with TGF␤ induced PI3 kinase activity as well as Smad2 and Akt phosphorylation (Fig. 4).
Active T␤RI (Alk5 T204D ) Can Induce PI3 Kinase Activity-Because p85 can bind T␤RII (Fig. 2), we next determined whether overexpression of T␤RII was permissive for ligandinduced PI3 kinase and P-Akt and compared these with the kinetics of activation observed in untransfected COS7 cells (Fig. 1). In COS7 cells overexpressing transfected T␤RII, TGF␤ maximally induced PI3 kinase activity and Ser-473 P-Akt at 6 h (Fig. 5) suggesting that an excess of type II receptor did not FIG. 2. p85 binds to T␤RI and T␤RII. COS7 cells were transiently transfected with vectors encoding p85 and T␤RI-HA or p85 and T␤RII-Flag. A, cells without TGF␤ treatment were lysed with a buffer containing 1% Nonidet P-40 and precipitated (IP) with T␤RI or T␤RII antibodies, respectively. Immunoprecipitates were subjected to a p85 immunoblot procedure (IB, top panel). Expression of the transfected proteins in whole cells lysates was checked by p85, HA, and FLAG immunoblot (3 bottom panels). The p85 band in the control (Ctl) lysate lanes represents endogenous protein. B, Twenty-four hours after transfection with vectors encoding p85 and T␤RI-HA or p85 and T␤RII-Flag, the cells were serum-starved for another 24 h and then treated or not treated with TGF␤ (2 ng/ml) for the indicated times. Cell lysates were precipitated with T␤RI or T␤RII antibodies, and the precipitates were immunoblotted with a p85 antibody (top panel only). Expression of the transfected p85, T␤RI, and T␤RII was detected by p85, HA, and FLAG immunoblot, respectively. Total Smad2, P-Smad2, total Akt, and P-Akt were also determined by immunoblot analyses as indicated to the right of each panel. C, 35 S-labeled p85 generated by in vitro transcription/translation in the presence of T7 polymerase was incubated with GST alone, GST-T␤RI, or GST-T␤RII as indicated under "Materials and Methods." Labeled p85 was separated from GSH beads by 8% SDS-PAGE and visualized by autoradiography. Sp6 polymerase was used as a negative control. Bottom panel shows the input of GST and GST-fusion proteins into the SDS gels. accelerate PI3 kinase activation compared with untransfected cells. We were unable to detect any PI3 kinase activity in immune complexes precipitated with T␤RII antibodies (not shown).
The studies above had shown that in cells overexpressing transfected p85 and T␤RI, the association of p85 and type I receptors was maximal after 1 h of treatment with TGF␤ (Fig.  2B). Moreover, in COS7 and R1B cells overexpressing transfected T␤RI, maximal ligand-induced activation of PI3 kinase was evident after 1 h of treatment (Figs. 3B and 4B), suggesting that T␤RI can accelerate the activation of PI3 kinase. Therefore, to test whether T␤RI can activate PI3 kinase, we transduced NMuMG and R1B cells with an HA-tagged adenovirus encoding a mutant of T␤RI (Alk5 T204D ). A substitution of threonine 204 with aspartic acid leads to constitutive activation of the type I receptor serine-threonine kinase thus allowing it to signal in the absence of added ligand (20). In both cells, transduction with a mutant Alk5 T204D but not with a ␤-galactosidase control adenovirus resulted in a marked increase in ligand-independent, p85-associated PI3 kinase activity as well as Smad2 phosphorylation (Fig. 6).
To determine whether active Alk5 was causally associated with the enhanced PI3 kinase activity observed, we utilized small molecule T␤RI serine-threonine kinase inhibitor LY580276 (21). LY580276 is a parafluorophenyl substituted dihydropyrrolopyrazole that is highly selective for T␤RI (IC 50 0.18 M, K i 37 nM) relative to T␤RII (IC 50 Ͼ20 M), p38␣ (IC 50 Ͼ20 M) and a panel of 39 additional kinases with an IC 50 Ͼ10 M. 2 It inhibits TGF␤-stimulated p3TP-Lux reporter activity in Mink lung cells (IC 50 96 nM) and NIH3T3 proliferation (IC 50 39 nM). The crystal structure of LY580276 in Alk5 T204D has been solved, which confirms binding at the ATP site (22). Treatment with LY580276 for 24 h blocked Alk5 T204D -mediated PI3 kinase activation and Smad2 phosphorylation (Fig. 6) supporting a causal association between active Alk5 signaling and the induction of PI3 kinase catalytic activity.
Smad7 Blocks Alk5-mediated Activation of PI3 Kinase-To determine whether the activation of PI3 kinase by T␤RI requires Smad function, we blocked Smad signaling with the inhibitory Smad7 (23,24). Smad7 stably interacts with activated ⌻␤RI thereby blocking the association with phosphorylation and activation of Smad2/3. Overexpressed adenoviral Smad7 blocked Alk5 T204D -induced PI3 kinase activity as well as Smad2 phosphorylation (Fig. 7). DISCUSSION We have examined whether TGF␤ receptors interact with and/or activate PI3 kinase in epithelial cells. Exogenous TGF␤ increased PI3 kinase activity as measured by the ability of p85-associated immune complexes to phosphorylate inositides in vitro. The kinetics of ligand-induced PI3 kinase activation were much slower than the phosphorylation of Smad2 with maximal detectable induction 6 h after the addition of TGF␤. Both type I and type II receptors associated with p85, but the association of T␤RII appeared constitutive, as it was not enhanced by exogenous ligand. On the other hand, the interaction of T␤RI with p85 was markedly enhanced by treatment with TGF␤ for 1 h. This receptor association with PI3 kinase was not direct. 35 S-Labeled rabbit reticulocyte p85 did not associate with fusion proteins containing type I and type II receptors. In addition, we were unable to detect any PI3 kinase activity in T␤RI and T␤RII precipitates from TGF␤-stimulated cells. This is consistent with a previous report using TGF␤-stimulated fibroblasts in which PI3 kinase activity was not recovered with a T␤RII antibody (8). These results and the slow kinetics of induction of PI3 kinase catalytic activity by TGF␤ suggest the presence of signaling intermediates between ligand-bound TGF␤ receptors and PI3 kinase. Although T␤RII overexpres-2 J. Yingling, personal communication.

FIG. 3. Kinase-inactive mutant of T␤RII inhibits the interaction between p85 and T␤RI and TGF␤-induced PI3 kinase activity.
A, COS7 cells were transiently transfected with p85 and T␤RI-HA with or without kinase-dead T␤RII K277R -HA. Twenty-four hours after transfection, the cells were serum-starved for another 24 h and then treated with or without TGF␤ (2 ng/ml) for 1 h. Cell lysates were precipitated (IP) with T␤RI or T␤RII antibodies and immunoblotted (IB) with a p85 antibody (top panel). Expression of the transfected proteins in whole cells lysates was detected by p85, HA (for T␤RI), and T␤RII (for T␤RII K277R ) immunoblot analyses (3 bottom panels). B, COS7 cells were transiently transfected with either T␤RI alone or T␤RI and T␤RII K277R . Cell lysates were collected after treatment with TGF␤ (2 ng/ml) for the indicated times and precipitated with a p85 antibody. Immune complexes were tested in a PI3 kinase reaction (left). Expression of the transfected receptor proteins in whole cells lysates was detected by HA and T␤RII immunoblots (right). Wn, wortmannin.
FIG. 4. T␤RI is required for TGF␤-induced PI3 kinase activity. R1B cells were stably transduced with retroviruses encoding T␤RI or vector alone. Transduced RIB cells were incubated with serum-free medium for 24 h and then treated with or without TGF␤ (2 ng/ml) for the indicated times. Cell lysates from vector control (A) and T␤RI-reconstituted cells (B) were collected and tested in a PI3 kinase in vitro kinase assay (left panels). T␤RI, total Smad2, P-Smad2, total Akt, and P-Akt were measured in whole cell lysates by immunoblot (right panels). Wn, wortmannin.
FIG. 5. Overexpression of T␤RII does not accelerate PI3 kinase activation. COS7 cells were transiently transfected with T␤RII-Flag. Twenty-four hours after transfection, the cells were serum-starved for 24 h and then treated with or without TGF␤ (2 ng/ml) for the indicated times. Cell lysates were collected, precipitated with a p85 antibody, and subjected to a PI3 kinase in vitro assay (left panel). Transfected T␤RII expression level was detected by FLAG immunoblot. Phosphorylation and content of Smad2, Akt, and MAPK were detected by immunoblot analysis of whole cell lysates (right panel). Wn, wortmannin.
sion did not increase TGF␤-induced or uninduced PI3 kinase activity, a kinase-dead, dominant-negative mutant of T␤RII blocked ligand-induced p85-⌻␤RI association and PI3 kinase activity suggesting that a functional type II receptor is required for TGF␤-mediated activation of PI3 kinase. These data also imply that in tumors with inactivating mutations in the TGFBR2 gene TGF␤ may not be able to engage PI3 kinase and contribute to tumor progression by this effector pathway.
In T␤RI-null R1B cells, TGF␤ did not stimulate PI3 kinase activity. However, this stimulation was restored upon reconstitution of T␤RI by transfection. Moreover, the overexpression of T␤RI accelerated ligand-induced catalytic activity with maximal induction observed 1 h after the addition of TGF␤. To confirm that the serine-threonine kinase activity of T␤RI can activate PI3 kinase, we transduced R1B and NMuMG epithelial cells with a dominant active mutant form of T␤RI. In both cell lines, the expression of T␤RI T204D markedly enhanced ligand-independent PI3 kinase activity, which was blocked by the addition of the T␤RI kinase inhibitor LY580276, strongly suggesting a causal link between T␤RI function and PI3 kinase. Finally, overexpressed Smad7, which binds phosphorylated T␤RI, prevented ligand-induced PI3 kinase activity as well as Smad2 phosphorylation, implying further that T␤RI can associate with and activate PI3 kinase.
Despite this causal evidence between T␤RI function and TGF␤-mediated activation of PI3 kinase, it is not clear how T␤RI stimulates this activity. The level of induction with added ligand is comparable with that reported in platelet-derived growth factor-stimulated Swiss 3T3 (8) and NIH3T3 cells (25) where the p85-associated PI3 kinase activity is only increased ϳ20% over untreated controls. Only 3-10% of cellular PI3 kinase is activated by insulin, colony-stimulating factor-1, or erbB2 receptors (26 -28). This level of activation probably reflects the large pool of PI3 kinase that is modestly affected by a single growth factor. In human airway smooth muscle cells, for example, TGF␤ can induce a detectable increase in PI3 kinase only in the presence of epidermal growth factor (29). Further consistent with this notion, overexpression of p85 in COS7 cells eliminated the detectable increase in p85-associated PI3 kinase activation in response to an added ligand (data not shown). Nonetheless, the marked induction of p85-associated PI3 kinase observed in cells transduced with the constitutively active T␤RI T204D mutant, which was blocked by a pharmacological inhibitor of the type I receptor kinase, pro-FIG. 6. Constitutively active T␤RI mutant induces PI3 kinase activity. NMuMG (A) and R1B (B) were transduced with (control) ␤-galactosidase (␤Gal) or Alk5 T204D -HA adenoviruses (3 multiplicity of infection each). Twenty-four hours after transduction, the small molecule inhibitor of the T␤RI kinase, LY580276 (10 M), was added. After a 24-h incubation, cell lysates were collected, precipitated with a p85 antibody, and tested in a PI3 kinase in vitro reaction (left panels). Alk5 T204D -HA, Smad2, and P-Smad2 levels were measured by immunoblot procedures using the indicated antibodies (right panels). Wn, wortmannin. vides causal evidence linking T␤RI and PI3 kinase activities. Similar results (17) have been reported in MDA-231 human breast cancer cells expressing kinase-dead T␤RII K277R : Transduction with an adenovirus encoding activated T␤RI induced P-Akt and restored PI3 kinase-dependent cell motility (17).
PI3 kinase is usually induced by activated transmembrane tyrosine kinases such as insulin (30), nerve growth factor (31), and platelet-derived growth factor receptors (25). Activated epidermal growth factor receptors (erbB1) or erbB2 can also link to PI3 kinase activity after partnering with erbB3, a coreceptor with six p85-binding sites (32)(33)(34). Upon receptor activation, p85 associates via its SH2 domain with phosphorylated tyrosines within YXXM motifs in the C terminus of the receptor. Because of this, phosphotyrosine antibody precipitates from platelet-derived growth factor-stimulated cells have been shown to contain 50-fold more PI3 kinase activity than unstimulated controls (35). YXXM motifs are not present in T␤RI or T␤RII, which is further in line with the lack of evidence for a direct association between p85 and TGF␤ receptors (Fig.  2C). In some cases, the interaction between p85 and receptor tyrosine kinases is also indirect and occurs through intermediate phosphoproteins such as that reported for the insulin receptor substrates IRS1 and IRS2 (36,37). We were unable to recover any detectable PI3 kinase activity in phosphotyrosine immunoprecipitates from TGF␤-treated cells (data not shown), suggesting that PI3 kinase activation is not mediated by tyrosine phosphorylation. Therefore, we suspect this is the first report of PI3 kinase activation induced by a serine-threonine kinase in epithelial cells.
It remains to be determined whether TGF␤-induced PI3 kinase activity is universal among epithelial cells. If in fact, as our data would suggest, PI3 kinase activation requires signaling intermediates, the differential expression of those putative intermediates may specify whether TGF␤ can or cannot activate this signaling pathway as well as the magnitude and time course of this activation. Indeed, we have not observed ligandinduced activation of Akt in MCF-10A and MCF-7 human mammary epithelial cells. Because Akt activity is PI3 kinasedependent, this result would suggest that activation of PI3 kinase by TGF␤ is cell type-dependent. The answer to these questions will require further investigation.