Cellular Conditioning and Activation of β-Catenin Signaling by the FPB Prostanoid Receptor*

FP prostanoid receptors have been identified as two isoforms named FPA and FPB. We have shown that the FPB isoform, but not the FPA, activates β-catenin-mediated transcription. We now report that the mechanism of this FPB-specific activation of β-catenin signaling occurs in two steps. The first is a conditioning step that involves an agonist-independent association of the FPBreceptor with phosphatidylinositol 3-kinase followed by constitutive internalization of a receptor complex containing E-cadherin and β-catenin. This constitutive internalization conditions the cell for subsequent β-catenin signaling by increasing the cellular content of cytosolic β-catenin. The second step involves agonist-dependent activation of Rho followed by cell rounding. Because of the conditioning step, this agonist-dependent step results in a stabilization of β-catenin and activation of transcription. Although stimulation of the FPA isoform activates Rho and induces cellular shape change, it does not activate β-catenin signaling, because the FPA does not undergo constitutive internalization and does not condition the cell for β-catenin signaling. The cellular conditioning described here for the FPB illustrates the potential of the receptor to alter the signaling environment of a cell even in the absence of agonist and has general significance for understanding G-protein-coupled receptor signaling.

Prostaglandin F 2␣ (PGF 2␣ ) 1 is an important autacoid that regulates a variety of physiological processes such as inflammation, cardiac hypertrophy, intraocular pressure, and regression of corpus luteum. PGF 2␣ is synthesized from arachidonic acid by the cyclooxygenases and binds to FP prostanoid receptors to initiate its signaling cascade. FP receptors are in the family of prostanoid receptors, which in turn are in the superfamily of G-protein-coupled receptors (GPCRs). Other prostanoid receptors include the EP, DP, IP, and TP receptors, which mediate, respectively, the actions of prostaglandin E 2 , prostaglandin D 2 , prostacyclin, and thromboxane.
The prostanoid receptor family also includes additional EP receptor subtypes (EP 1 , EP 2 , EP 3 , and EP 4 ) and alternative mRNA splice variants of the EP 1 , EP 3 , TP, and FP receptors. The FP receptor splice variants are designated FP A and FP B and are identical except for their intracellular carboxyl-terminal domains. Thus, the FP B isoform is basically a truncated version of the FP A isoform, which lacks the last 46 carboxylterminal amino acids. Both receptor isoforms are coupled to G q and can activate phosphatidylinositol signaling followed by the activation of protein kinase C. In addition both isoforms activate Rho, a member of the Ras family of small GTPases. The activation of Rho by FP A and FP B receptor isoforms leads to tyrosine phosphorylation of p125 focal adhesion kinase and the induction of cellular shape change involving the retraction of filopodia, cell rounding, and aggregation (1). Unexpectedly the reversal of this Rho-mediated shape change was found to differ markedly between the two isoforms following the removal of agonist (2). Thus, 1 h after the removal of PGF 2␣ , FP A -expressing cells return to their original cellular morphology, whereas FP B -expressing cells remain rounded and still show evidence of Rho-mediated signaling, including the presence of actin stress fibers and tyrosine phosphorylation of p125 focal adhesion kinase. Even 16 h after the removal of PGF 2␣ , FP B -expressing cells are still rounded. In addition we have found that stimulation of FP B -expressing cells with PGF 2␣ produces a marked activation of Tcf/␤-catenin-mediated transcriptional activation, which is not observed in FP A -expressing cells (3).
We have recently documented another fundamental difference between these FP receptor isoforms (4). We have found that the FP A isoform undergoes a classic agonist-induced and clathrin-dependent internalization, whereas the FP B isoform undergoes an agonist-independent constitutive internalization that does not involve clathrin. We now report a molecular mechanism that we believe links the constitutive agonist-independent internalization of the FP B isoform with the selective activation of Tcf/␤-catenin signaling by the FP B isoform. A key observation is that agonist-induced Tcf/␤-catenin transcriptional activation by the FP B isoform was blocked by inhibition of Rho-mediated cellular shape change. Furthermore, we have found an association of the FP B isoform, but not the FP A isoform, with phosphatidylinositol 3-kinase (PI3K), which may explain its agonist-independent constitutive internalization. We hypothesize that constitutive internalization of the FP B isoform involving PI3K conditions FP B -expressing cells to subsequent agonist-induced Tcf/␤-catenin signaling by increasing the cellular content of cytosolic ␤-catenin.

EXPERIMENTAL PROCEDURES
Immunoprecipitation and Western Blotting-HEK-293 cells stably expressing FP A and FP B prostanoid receptor isoforms (1), as well as FLAG-tagged FP A -and FP B -expressing cell lines (5), were generated and cultured as described previously. Cells were pretreated with either vehicle (0.1% Me 2 SO or water), or 100 nM wortmannin (Sigma) for 15 min, or 40 g/ml C3-toxin for 48 h at 37°C. Then cells were incubated at 37°C with either vehicle (sodium carbonate, 0.002% final) or 1 M PGF 2␣ (Cayman Chemical) for the times indicated in the figures. Cells were scraped and sonicated in a lysis buffer as described previously (3). Samples were centrifuged, the supernatant (cytosolic fraction) was removed, and the pellet (particulate fraction) was solubilized with lysis buffer containing 0.2% Triton X-100 (Bio-Rad) and then centrifuged again to remove insoluble debris. For immunoprecipitation (IP), samples (200 -300 g of protein) were rotated at 4°C with anti-FLAG M2 affinity gel (Sigma) for 2 h, or with anti-E-cadherin antibody (BD Transduction Laboratories) for 16 h, or with anti-␤-catenin antibody (BD Transduction Laboratories) for 2 h, all at a dilution of 1:100. For the E-cadherin and ␤-catenin IPs, 10 l of a 1:1 slurry of protein G-Sepharose (Amersham Biosciences) was added and rotated for another hour. Samples were electrophoresed, transferred to nitrocellulose membranes, and incubated in 3% nonfat milk with 1:1,000 dilutions of either anti-PI3K p85 antibody (Upstate Biotechnology Inc.) for 2 h or a mixture of mouse monoclonal antibodies to phospho-serine (Sigma) and phospho-threonine (Sigma) for 16 h (3). The membranes were washed and incubated for 1 h at room temperature in 3% nonfat milk containing 1:10,000 dilutions of horseradish peroxidase-conjugated goat anti-rabbit (Sigma) or anti-mouse antibodies (Sigma). The immunoblots were then visualized by enhanced chemiluminescence (SuperSignal, Pierce). To verify amounts of immunoprecipitated proteins, the membranes were stripped and reprobed with 1:10,000 dilutions of either anti-FLAG M2 antibody (Sigma), anti-E-cadherin, or anti-␤-catenin antibody for 16 h in 3% nonfat milk under the same conditions as described above. For the immunoprecipitations of phospho-GSK-3␤, cells were scraped into a lysis buffer as described previously (6), and 100 g of protein was electrophoresed then transferred to nitrocellulose membranes. Membranes were incubated in 5% nonfat milk for 1 h and were washed and incubated for 16 h at 4°C in 0.5% nonfat milk containing either antiphospho-GSK-3␤ antibody (#9336, Cell Signaling) or anti-GSK-3␤ antibody (G22320, Transduction Laboratories) as previously described (6). All antibodies were used at a dilution of 1:1,000. Membranes were washed three times and incubated for 1 h at room temperature in 0.5% nonfat milk for GSK-3␤ antibodies and then with a 1:10,000 dilution of the corresponding secondary antibodies conjugated with horseradish peroxidase. To ensure equal loading of proteins, the membranes were stripped and reprobed with anti-GSK-3␤ antibodies under the same conditions as described above. The resulting films were scanned, and quantitation was performed as described previously (3).
PI3K Activity Assay-Cells were pretreated with either vehicle (0.1% Me 2 SO) or 100 nM wortmannin for 15 min at 37°C followed by stimulation with either vehicle (0.002% sodium carbonate) or 1 M PGF 2␣ for 1 h. Cells were then scraped in a lysis buffer consisting of 20 mM Tris-HCl (pH 7.5), 137 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10% (v/v) glycerol, 1% Nonidet P-40, 1 mM sodium orthovanadate, and 1 mM phenylmethylsulfonyl fluoride and rotated for 20 min at 4°C. For immunoprecipitation, samples were rotated with 500 g of protein with anti-PI3K p85 for 2 h followed by the addition of 10 l of 1:1 slurry of protein A-Sepharose (Sigma) for an additional hour. Samples were then washed twice with lysis buffer and twice with wash buffer consisting of 10 mM Tris-HCl (pH 7.5), 100 mM NaCl, and 1 mM EDTA followed by incubation with 0.1 mg/ml L-␣-phosphatidylinositol (Avanti polar lipids) in buffer containing 880 M cold ATP with 30 Ci of [␥-32 P]ATP (Amersham Biosciences), and 20 mM MgCl 2 for 10 min at 30°C in a final volume of 80 l. To terminate the incubation, 20 l of 8 M HCl was added to the samples, and then the lipids were extracted with 160 l of chloroform/methanol (1:1) and centrifuged (300 ϫ g) for 15 min at 4°C. The lower organic phases were removed and applied to an LK6D silica gel 60-Å TLC plate (Whatman). TLC plates were developed in chloroform/methanol/water/ammonium hydroxide (30:23.5:5.65:1, v/v) followed by drying and visualization by autoradiography using Hyperfilm MP (Amersham Biosciences).
Whole Cell Radioligand Binding-Cells were pretreated with either vehicle (0.1% Me 2 SO) or 100 nM wortmannin for 15 min at 37°C followed by stimulation with either vehicle (0.002% sodium carbonate) or 1 M PGF 2␣ for 1 h. They were then trypsinized, centrifuged at 500 ϫ g for 2 min, and resuspended at a concentration of 10 7 cells/ml in ice-cold MES buffer (2). [ 3 H]PGF 2␣ binding was performed using 2.5 nM [ 3 H]PGF 2␣ (Amersham Biosciences) as previously described (2). Samples were incubated for 1 h at room temperature, and the assays were terminated by filtration through Whatman GF/C glass filters using a cell harvester (M-24R, Brandel) (2).
Whole Cell Immunofluorescence Labeling of FP Receptors-Cells were split and grown in six-well plates containing 22-mm round glass coverslips for 3-4 days. To evaluate agonist-dependent internalization, the cells were pretreated with either vehicle (0.1% Me 2 SO) or 100 nM wortmannin for 15 min at 37°C followed by the addition of a 1:500 dilution of anti-FLAG M2 antibodies in vehicle (0.002% sodium carbonate) or a 1:500 dilution of anti-FLAG M2 antibodies in 1 M PGF 2␣ for 10 min at 37°C. The cells were fixed, permeabilized, and labeled with fluorescein isothiocyanate-conjugated anti-mouse IgG secondary antibodies as previously described (4). The cells were then examined by scanning confocal microscopy as described previously (4).
Immunofluorescence Labeling of ␤-Catenin-Cells were split and grown in six-well plates containing 22-mm round glass cover slips for 3-4 days. Cells were pretreated with either vehicle (0.1% Me 2 SO or water) or inhibitors, 100 nM wortmannin for 15 min, or 40 g/ml C3-toxin for 48 h at 37°C. Cells were then incubated at 37°C with either vehicle (0.002% sodium carbonate) or 1 M PGF 2␣ for 1 h and were rapidly washed and fixed with methanol/acetone (7:3, v/v). The fixed cells were incubated with a 1:1000 dilution of antibody to ␤-catenin in 3% bovine serum albumin and were then washed and incubated with a 1:10,000 dilution of fluorescein isothiocyanate-conjugated goat anti-mouse secondary antibody in 0.1% nonfat milk. Nuclei were stained with 0.05 nM/ml 4Ј,6-diamidino-2-phenylindole (DAPI, Sigma). Cells were visualized by phase-contrast microscopy, as described previously (3).
Tcf/Lef Reporter Gene Experiments-Cells, grown in six-well plates, were transiently transfected using FuGENE 6 (Roche Molecular Biochemicals) with 1 g/well of either the TOP flash or FOP flash reporter plasmids (Upstate Biotechnology Inc.) as described previously (3). Cells were pretreated with either vehicle (0.1% Me 2 SO or water), 100 nM wortmannin for 15 min, or 40 g/ml C3-toxin for 48 h at 37°C. The cells were then incubated at 37°C with either vehicle (0.002% sodium carbonate) or 1 M PGF 2␣ for 1 h and were rapidly washed three times each with 1 ml/well Opti-MEM and then incubated for 16 h at 37°C in 2 ml of Opti-MEM containing 250 g/ml Geneticin, 100 g/ml gentamicin. Cell extracts were prepared using the Luciferase assay system (Promega). Luciferase activity was measured using a Turner TD-20/20 luminometer as described previously (3) using 10 g of protein per sample. Measurements were corrected for background activity by subtraction of the FOP flash values from the corresponding TOP flash values.
[ 3 H]Inositol Phosphate Experiments-Cells were cultured in 6-cm plates in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and pretreated either with vehicle (water) or 40 g/ml C3-toxin for 48 h at 37°C followed by preincubation with 3 Ci/ml myo-[2-3 H]inositol (Amersham Biosciences) for 16 h. Cells were then incubated at 37°C with either vehicle (0.002% sodium carbonate) or 1 M PGF 2␣ for 1 h at 37°C in culture media containing 10 mM LiCl. Assays were terminated by the addition of 1 ml of methanol, and cells were scraped and added to 1.5 ml of chloroform/water (1:0.5, v/v). Total [ 3 H]inositol phosphates were then determined by anion-exchange chromatography as described previously (5).

Interaction of PI3K with the FP B Prostanoid Receptor-We
have previously shown that PGF 2␣ can activate Tcf/␤-catenin signaling in FP B -expressing cells but not in FP A -expressing cells (3). This selective activation of Tcf/␤-catenin signaling by the FP B isoform was associated with an agonist-mediated increase in cytoplasmic ␤-catenin and a decrease in its phosphorylation status. Decreased phosphorylation stabilizes ␤-catenin levels by preventing its degradation and thereby promotes nuclear translocation and Tcf transcriptional activation. However, the molecular mechanisms leading to the decrease in ␤-catenin phosphorylation following stimulation of the FP B receptor were unclear.
One possible mechanism leading to decreased phosphorylation of ␤-catenin would involve the sequential activation of PI3K and Akt resulting in the phosphorylation of glycogen synthase kinase-3 (GSK-3). Phosphorylation of GSK-3 inhibits its kinase activity and decreases ␤-catenin phosphorylation. The potential involvement of PI3K with FP B -mediated signaling was, therefore, explored in HEK cells stably expressing FLAG-tagged FP A and FP B receptors that were treated with 1 M PGF 2␣ and then examined by a combination of immunoblotting with antibodies to p85 PI3K (Fig. 1, A and B) and meas-urement of PI3K activity (Fig. 1C). FLAG-tagged FP-expressing cell lines were selected as reported previously (5) that had comparable levels of expression based on agonist-stimulated IP formation and on the radioligand binding (e.g. FP A , 3.56 Ϯ 0.04 pmol/mg of protein; FP B , 2.15 Ϯ 0.17 pmol/mg of protein). For the immunoblotting experiments the particulate fractions of cell lysates were first immunoprecipitated with monoclonal antibodies (M2) to the FLAG epitope. As shown in the upper panels of Fig. 1A, the p85 subunit of PI3K co-immunoprecipitated with the FP B receptor isoform but not with the FP A isoform. Interestingly the association of PI3K with the FP B isoform was disrupted by treatment with PGF 2␣ indicating a preferential interaction with the unstimulated receptor. The lower panels of Fig. 1A show that similar amounts of the FLAG-tagged FP A and FP B receptors were expressed in these cells and could be immunoprecipitated with the anti-FLAG M2 antibodies.
We then used wortmannin, an inhibitor of PI3K, to see if the association of the p85 subunit with the FP B receptor was influenced by its kinase activity. As shown in Fig. 1B, pretreatment of FP B -expressing cells with 100 nM wortmannin for 15 min significantly decreased the co-immunoprecipitation of the p85 subunit with the FP B receptor in the absence of PGF 2␣ treatment. As in Fig. 1A, treatment with 1 M PGF 2␣ for 1 h decreased the association of PI3K with the FP B receptor both under control conditions and following pretreatment with wortmannin.
PI3K activity was then determined under the same conditions as those used for the experiments in Fig. 1B, and the results are shown in Fig. 1C. PI3K activity was greater in FP B -expressing cells as compared with FP A -expressing cells and was decreased ϳ50% in FP B -expressing cells after treatment with 1 M PGF 2␣ for 1 h. As expected, pretreatment with wortmannin decreased the PI3K activity in both vehicletreated FP B cells and in cells treated with 1 M PGF 2␣ for 1 h. In summary, it appears that PI3K activity is required for an interaction with the FP B receptor and may, in fact, be induced by this interaction. This is supported by the observation that treatment of FP B -expressing cells with PGF 2␣ decreases PI3K activity while at the same time decreasing its interaction with the receptor.
Inhibition of PI3K Inhibits Agonist-independent Internalization of FP B Receptors-It has been reported that the endocytosis of E-cadherin is regulated by Rac1 and that this may involve the activation of PI3K (7,8). Furthermore, it appears that the endocytosis of E-cadherin involves a clathrin-independent mechanism (9). Recently we have found that the FP B receptor isoform undergoes an agonist-independent constitutive inter- nalization that is also clathrin-independent, whereas, the FP A isoform undergoes agonist-induced internalization that is clathrin-dependent (4). Given the association of PI3K with the FP B isoform (shown in Fig. 1) we were interested in the potential role of PI3K with respect to the agonist-independent constitutive internalization of the FP B isoform. For these experiments HEK cells stably expressing the FP A and FP B receptor isoforms were pretreated with either vehicle or 100 nM wortmannin followed by treatment with either vehicle or 1 M PGF 2␣ . Cell lines expressing wild type FP receptors were selected as reported previously (5) that had comparable levels of expression based on agonist-stimulated IP formation and on the radioligand binding (e.g. FP A , 3.55 Ϯ 0.28 pmol/mg of protein; FP B , 4.09 Ϯ 0.49 pmol/mg of protein). Receptor desensitization and internalization were then assessed, respectively, by the whole cell binding of [ 3 H]PGF 2␣ ( Fig. 2A) and by immunofluorescence confocal microscopy with anti-FLAG M2 antibodies (Fig. 2B).
As shown in Fig. 2A, treatment with PGF 2␣ resulted in a 50% decrease in [ 3 H]PGF 2␣ binding in both FP A -and FP B -expressing cells (2). Interestingly pretreatment with wortmannin, alone, increased [ 3 H]PGF 2␣ binding in FP B cells, but not in FP A cells. This sensitization of [ 3 H]PGF 2␣ binding following wortmannin pretreatment, although slight, was statistically significant and was obtained consistently and repeatedly in FP B cells, but not in FP A cells. Fig. 2B shows the results of immunofluorescence microscopy of the FLAG-tagged FP A and FP B receptor isoforms using live cell labeling of cell surface receptors (4). A comparison of panels a and b shows that the FP A isoform is localized primarily on the cell surface membrane in vehicle-treated FP A cells and that following treatment with PGF 2␣ the receptor undergoes extensive internalization. A comparison of panels c and d shows that pretreatment of FP A cells with wortmannin did not affect this pattern of receptor localization. In contrast, examination of panels e and f shows that the FP B isoform is localized both intracellularly and on the cell surface regardless of PGF 2␣ treatment. This, as we have previously reported, reflects constitutive agonist-independent internalization of the FP B isoform. Comparison of panels e and g further shows that pretreatment with wortmannin reduced the intracellular localization of the FP B isoform in vehicletreated FP B -expressing cells; this suggests that inhibition of PI3K blocked the constitutive agonist-independent internalization of the FP B isoform. This is consistent with the sensitization of [ 3 H]PGF 2␣ binding observed above following pretreatment of the FP B cells with wortmannin.
Inhibition of PI3K Increases Membrane Association of E-Cadherin, ␤-Catenin, and PI3K-The interaction of E-cadherin with ␤-catenin is well established (10), and recent studies have shown additional interaction between E-cadherin and PI3K (8).
Having established an association between PI3K and the FP B receptor (Fig. 1) and an effect of wortmannin on FP B receptor localization (Fig. 2), we sought to determine the effects of wortmannin on the localization of E-cadherin and ␤-catenin in FP B -expressing HEK cells. Fig. 3 shows representative immunoblots (A) and the pooled densitometric analyses (B) for the expression of membrane associated (particulate) E-cadherin and ␤-catenin and for the expression of cytosolic ␤-catenin following pretreatment of FP A -and FP B -expressing cells with 100 nM wortmannin for 15 min. As we have previously reported, the expression of ␤-catenin is higher in FP B -expressing cells as compared with FP A -expressing cells in both the particulate and cytosolic fractions (3). We now show that the expression of particulate E-cadherin is also higher in FP B -expressing cells. Fig. 3B shows that wortmannin pretreatment of vehicletreated FP B cells slightly, but significantly, increased the ex-pression of membrane-associated (particulate) E-cadherin and ␤-catenin, while simultaneously decreasing the expression of cytosolic ␤-catenin. These increases in membrane-associated E-cadherin and ␤-catenin are very similar to the sensitization of [ 3 H]PGF 2␣ binding to the FP B receptors observed in Fig. 2A  following pretreatment with the PI3K inhibitor, wortmannin.
As noted above, PI3K has recently been found to bind Ecadherin (8). The immunoblot shown in Fig. 3C confirms this interaction by showing co-immunoprecipitation of the p85 subunit of PI3K with E-cadherin in both FP A -and FP B -expressing HEK cells. This figure also shows that PGF 2␣ treatment of FP B -expressing cells increased the co-immunoprecipitation of the p85 subunit with E-cadherin suggesting that PGF 2␣ treatment, which we have shown in Fig. 1C to inhibit PI3K activity, increases its association with E-cadherin. This conclusion is corroborated by the effects of the wortmannin pretreatment of vehicle-treated FP B cells, which also increased the co-immunoprecipitation of the p85 subunit with E-cadherin. Notably, such corresponding changes were not observed in the FP A -expressing cells.

Activation of PI3K Is Not Sufficient to Explain PGF 2␣ Stimulation of Tcf/␤-Catenin Signaling by the FP B Receptor-One
of the most pronounced effects of the treatment of FP B -expressing cells with PGF 2␣ is a dramatic reorganization of the ␤-catenin as evidenced by immunofluorescence microscopy using antibodies to ␤-catenin (3). Fig. 4 shows this effect in which the localization of ␤-catenin is green and the nuclear staining of 4Ј,6-diamidino-2-phenylindole (DAPI) is blue. Thus, a compar-ison of panels e and f shows that treatment with PGF 2␣ causes a marked increase in ␤-catenin immunofluorescence in regions of cell to cell contact. A comparison of panels a and b shows that a corresponding reorganization of ␤-catenin does not occur in FP A -expressing cells after treatment with PGF 2␣ . Pretreatment with wortmannin was used to determine the possible influence of PI3K activity on this PGF 2␣ -induced reorganization of ␤-catenin. Comparison of panels e and g shows that wortmannin pretreatment of vehicle-treated FP B cells produced a slight increase in ␤-catenin immunofluorescence in regions of cell to cell adhesion. However, as shown by comparing panels g and h, wortmannin pretreatment of FP B -expressing cells did not have any major effect on the PGF 2␣ -induced reorganization of ␤-catenin.
Stabilization of ␤-catenin expression and the promotion of Tcf transcriptional activation involves the decreased phosphorylation of ␤-catenin and is directly influenced by the activity of GSK-3␤. The activity of GSK-3␤, in turn, is a function of its state of phosphorylation, and it is known that phosphorylation of GSK-3␤ at serine 9 inhibits its kinase activity (11). Immunoblotting of phospho-GSK-3␤ and phospho-␤-catenin (Fig. 5A) and direct measurement of Tcf transcriptional activation (Fig.  5B) following pretreatment of FP B -expressing cells with wort- FIG. 3. Expression of E-cadherin  and ␤-catenin (A and B) and interaction of E-cadherin and PI3K (C). A, FP A -and FP B -expressing cells were pretreated with either vehicle or 100 nM wortmannin (wort) for 15 min followed by treatment with either vehicle (v) or 1 M PGF 2␣ (P) for 1 h. Particulate and cytosolic fractions were prepared as described under "Experimental Procedures" and were immunoblotted (IB) for E-cadherin (E-cad) and ␤-catenin (␤-cat). Shown is a representative immunoblot. B, histographs for the expression of E-cadherin and ␤-catenin as assessed by the pooled densitometry data (means Ϯ S.E.) from three independent experiments that were performed as in A. *, p Ͻ 0.05, as compared with the vehicle-treated FP B -expressing cells. C, same treatments as described in A followed first by immunoprecipitation (IP) with antibodies to E-cadherin (Ecad) and then immunoblotting (IB) with antibodies to either PI3K p85 (PI3K p85) or E-cadherin. Shown is a representative immunoblot and a histograph of the co-immunoprecipitation of PI3K p85 with E-cadherin as assessed by the pooled densitometry data (means Ϯ S.E.) from three independent experiments. Data are normalized to the vehicle-treated FP B -expressing cells as 100%. mannin were, therefore, used to assess the potential role of PI3K on the PGF 2␣ -induced activation of Tcf/␤-catenin signaling.
As shown in Fig. 5A, treatment of FP B -expressing cells with 1 M PGF 2␣ for 1 h increased the phosphorylation of GSK-3␤ (panel a) and was accompanied by a marked reduction in phosphorylation of cytosolic ␤-catenin (panel c). On the other hand, in FP A -expressing cells treatment with PGF 2␣ had no apparent effect on the phosphorylation of GSK-3␤ (panel a), and there was a marked increase in the phosphorylation of cytosolic ␤-catenin (panel c). It would be expected, therefore, that following treatment with PGF 2␣ the degradation of ␤-catenin would be favored in FP A -expressing cells, whereas, the stabilization of ␤-catenin and potential activation of transcription would be favored in FP B -expressing cells. In fact, as shown in Fig. 5B, this is what was found when transcriptional activation was measured with the use of the Tcf-responsive luciferase reporter gene. Thus, luciferase activity was stimulated ϳ3-fold in PGF 2␣ -treated FP B -expressing cells, whereas it was essentially unaffected in FP A -expressing cells as shown previously (3). Fig. 5 also shows that pretreatment with wortmannin reduced the basal levels of GSK-3␤ phosphorylation in both the vehicle-treated FP A -and FP B -expressing cells. As one might expect, there were corresponding increases in the phosphorylation of cytosolic ␤-catenin in both the vehicle-treated FP A -and FP B -expressing cells that were pretreated with wortmannin. When these same cells were treated with PGF 2␣ there were slight increases in the phosphorylation of GSK-3␤ in both FP A -and FP B -expressing cells and, correspondingly, slight decreases in the phosphorylation of ␤-catenin. Surprisingly, however, the Tcf-responsive reporter gene activity was only modestly inhibited (ϳ17%) in wortmannin-pretreated FP B -expressing cells that were treated with PGF 2␣ . This indicates that PGF 2␣ -induced activation of Tcf/␤-catenin signaling by the FP B isoform cannot be simply explained by direct activation of a PI3K/GSK-3␤ pathway.
Activation of Rho Is Required for PGF 2␣ Stimulation of Tcf/ ␤-catenin Signaling by the FP B Receptor-Because the activation of PI3K alone was not sufficient to explain the stimulation of Tcf/␤-catenin signaling by the FP B receptor, we decided to examine the influence of Rho. We have previously established that activation of Rho by PGF 2␣ in FP A -and FP B -expressing cells leads to morphological changes consisting of the loss of filopodia, cell rounding, and the formation of cell aggregates  (1). These morphological changes are reversible in FP A -expressing cells following the removal of PGF 2␣ , but, remarkably, they are not reversible in FP B -expressing cells (2). However, these morphological changes can be blocked in both FP A -and FP B -expressing cells by pretreatment with C3-toxin, a bacterial exoenzyme that ADP-ribosylates and inactivates Rho. In this series of experiments we used C3-toxin to examine its effects on PGF 2␣ -stimulated inositol phosphates formation (Fig. 6), cellular reorganization of ␤-catenin immunofluorescence (Fig. 7), phosphorylation of GSK-3␤ and ␤-catenin (Fig.  8A), and Tcf-responsive reporter gene activity (Fig. 8B) in HEK cells stably expressing the FP A and FP B receptor isoforms. Fig. 6A shows phase-contrast microscopy of FP A -and FP Bexpressing cells that were pretreated with either vehicle or 40 g/ml C3-toxin for 48 h followed by treatment with either vehicle or 1 M PGF 2␣ for 1 h. As we have previously reported (1), pretreatment with C3-toxin completely blocked PGF 2␣ -induced shape change in both FP A -expressing cells (compare panels b and d) and in FP B -expressing cells (compare panels f and h). Fig. 6B further shows that pretreatment with C3-toxin did not inhibit the PGF 2␣ -stimulated formation of inositol phosphates in either the FP A -expressing cells or the FP B -expressing cells. In fact, pretreatment with C3-toxin appears to have slightly enhanced the formation of inositol phosphates, indicating that the effects of C3-toxin on PGF 2␣ -induced shape changes were not a consequence of nonspecific cellular toxicity. Fig. 7 shows the immunofluorescence localization of ␤-catenin (green) and the nuclear staining of DAPI (blue) in FP A -and FP B -expressing HEK cells that were pretreated as above with either vehicle or C3-toxin and then with PGF 2␣ . Comparison of panels e and f shows that treatment with PGF 2␣ caused an increase in ␤-catenin immunofluorescence at regions of cell to cell adhesion in FP B -expressing cells that were not pretreated with C3-toxin. These findings are essentially the same as those shown in Fig. 4 for the corresponding experiments done with wortmannin. However, in contrast to the results obtained with wortmannin, a comparison of panels g and h of Fig. 7 shows that pretreatment of FP B -expressing cells with C3-toxin greatly reduced the PGF 2␣ -stimulated accumulation of ␤-catenin immunofluorescence at regions of cell to cell contact. PGF 2␣ treatment had no obvious effects on the immunolocalization of ␤-catenin in FP A -expressing cells that were pretreated with either vehicle (panels a and b) or C3-toxin (panels c and d). Fig. 8A shows the immunoblots that were obtained for phospho-GSK-3␤ (panel a) and for phospho-␤-catenin (panel c) following PGF 2␣ treatment of FP A -and FP B -expressing HEK cells that were pretreated with either vehicle or 40 g/ml C3-toxin for 48 h. As compared with the effects of wortmannin, which changed the phosphorylation patterns of GSK-3␤ and ␤-catenin (Fig. 5A), pretreatment with C3-toxin did not appear to significantly alter the phosphorylation of GSK-3␤ or ␤-catenin either under the basal conditions (vehicle treatment) or after treatment with 1 M PGF 2␣ for 1 h. Notably, pretreatment of FP Bexpressing cells with C3-toxin did not affect the dramatic PGF 2␣ -induced decrease in ␤-catenin phosphorylation as compared with FP B -expressing cells that were pretreated with vehicle (panel c). One might expect, therefore, that PGF 2␣induced activation of Tcf/␤-catenin signaling in FP B -expressing cells would be unaffected by pretreatment with C3-toxin. Fig.  8B, however, shows that C3-toxin pretreatment of FP B -expressing cells almost totally abolished PGF 2␣ -stimulated Tcfresponsive reporter gene activity. These results clearly establish the importance of Rho activation, or possibly shape change itself, for the PGF 2␣ stimulation of Tcf/␤-catenin signaling in cells expressing the FP B receptor. It may be recalled that pretreatment with C3-toxin did not block PGF 2␣ -stimulated inositol phosphates formation (Fig. 6B), therefore, these effects of C3-toxin on Tcf/␤-catenin signaling are unlikely to represent nonspecific cellular toxicity. DISCUSSION In this report we have explored the molecular mechanisms that are involved in the selective activation of Tcf/␤-catenin signaling by the FP B prostanoid receptor isoform. We have several major findings. The first is that the FP B receptor can interact with the p85 subunit of PI3K and both the activation of PI3K and its interaction with the receptor are disrupted in the presence of PGF 2␣ . The second finding is that inhibition of PI3K inhibits the constitutive, agonist-independent, internalization of the FP B receptor and this is associated with an accumulation of the FP B receptor on the cell surface membrane. Inhibition of PI3K activity is also accompanied by an increase in the membrane-associated expression of E-cadherin, Step 1 (Conditioning): In the absence of PGF 2␣ there is an association of PI3K with the FP B receptor that leads to activation of the kinase. This activation of PI3K leads to a constitutive (agonist-independent) internalization of the FP B receptor in a membrane complex containing E-cadherin and ␤-catenin. The escape of ␤-catenin from this complex increases the concentration of cytosolic ␤-catenin.
Step 2 (Activation): In the presence of PGF 2␣ , activation of the FP B receptor leads to activation of the protein kinase C and Rho signaling pathways. In addition, PI3K dissociates from the receptor leading to a decrease of kinase activity. The activation of the Rho signaling pathway leads to cell rounding and a dramatic increase of E-cadherin and ␤-catenin in the region of cell-cell adhesion. This morphological and molecular reorganization further increases the concentration of cytosolic ␤-catenin to a point that exceeds the capacity of the cell to degrade it leading to Tcf transcriptional activation. This does not occur in FP A -expressing cells, because they lack the conditioning step and have a lower initial concentration of cytoplasmic ␤-catenin. ␤-catenin, and PI3K. As illustrated in Fig. 9 these findings support the notion of a macromolecular complex involving the FP B receptor, PI3K, E-cadherin, and ␤-catenin. This complex undergoes constitutive internalization in the absence of PGF 2␣ and can be blocked when PI3K activity is inhibited with wortmannin. Although we could not immunoprecipitate E-cadherin or ␤-catenin with FP B receptor (data not shown), this might be explained if the complex is not tightly associated and especially if PI3K functions as a scaffold between the receptor and E-cadherin.
Two other major findings are as follows. First, a PI3K pathway, alone, is not sufficient to explain the PGF 2␣ -induced stimulation of Tcf transcriptional activation by the FP B receptor. The second is that activation of Rho is clearly required for FP B receptor-mediated Tcf transcriptional activation. There are several additional observations to note in terms of developing a molecular mechanism to explain the selective activation of Tcf/␤-catenin signaling by the FP B receptor. One is that the expression of both particulate and cytosolic E-cadherin and ␤-catenin are significantly greater in FP B -expressing cells as compared with FP A -expressing cells. We have previously observed that PGF 2␣ stimulation of FP B -expressing cells, but not FP A -expressing cells, causes a major reorganization of ␤-catenin in which there is an increased accumulation of ␤-catenin along regions of cell to cell contact (3). In addition, PGF 2␣ stimulation of the FP A receptor does not activate Tcf/␤-catenin signaling, nor does it induce the reorganization of ␤-catenin, even though it activates Rho and induces cell rounding (1).
To explain these findings we propose a two step mechanism in which FP B -expressing cells first undergo an agonist-independent conditioning step involving constitutive internalization of the FP B receptor followed by an agonist-dependent activation step leading to Tcf transcriptional activation. FP Aexpressing cells fail to activate Tcf/␤-catenin signaling, because they lack the initial conditioning step. This two-step mechanism is illustrated in more detail in Fig. 9. In support of the conditioning step we have previously shown agonist-independent constitutive internalization of the FP B receptor, but not the FP A receptor (4). Furthermore, this constitutive internalization of the FP B was found to be clathrin-independent. It is also well established that E-cadherin can associate with ␤-catenin, and recently it has been shown that the endocytosis of E-cadherin is clathrin-independent (9) and involves the activation of PI3K (7,8). In this study we have found that the FP B receptor isoform, but not the FP A , can associate with PI3K. We hypothesize that this interaction results in the activation of PI3K and thereby promotes binding to E-cadherin, which leads to the internalization of this protein complex. In other words, the constitutive internalization of the FP B receptor is a PI3K/ E-cadherin-dependent, but clathrin-independent process. Further consequences of this constitutive internalization is an increase in cytosolic ␤-catenin, which we have documented, as well as an increase in the basal level of agonist-independent Tcf transcriptional activation (3) (see also Figs. 5B and 8B). Therefore as compared with FP A -expressing cells, FP B -expressing cells have increased levels of cellular E-cadherin and ␤-catenin and are likely to have more active cycling of these proteins between the cell membrane and cytosolic compartments.
The activation step is initiated through the specific binding of PGF 2␣ to the FP B receptor. The immediate consequence of this is stimulation of phosphatidylinositol and Rho signaling as well as dissociation of PI3K from the FP B receptor and a decrease in cellular PI3K activity. In both FP A -and FP B -expressing cells this is followed by cell rounding and aggregation. However, in FP B cells, but not FP A cells, there is a critical accumulation of E-cadherin and ␤-catenin at regions of cell to cell contact that occurs simultaneously with cell rounding and aggregation. Similarly in FP B -expressing cells, but not in FP Aexpressing cells, there is stimulation of Tcf transcriptional activation. It is clear that there is an absolute requirement for stimulation of the FP B receptor. Thus, cell rounding and aggregation can be induced in FP B -expressing cells by treatment with lysophosphatidic acid, but these effects are reversible and do not activate Tcf/␤-catenin signaling (2). 2 The increased levels of E-cadherin and ␤-catenin that occur during the conditioning step also appear to be absolute requirements for the agonist-induced activation of Tcf/␤-catenin signaling in FP B -expressing cells. It would be expected that the loss of filopodia, cell rounding, and aggregation would dramatically decrease cellular surface area and essentially concentrate and further increase the functional consequences of the increased expression of E-cadherin and ␤-catenin. For reasons that we still do not fully appreciate, the cell rounding, aggregation, and reorganization of ␤-catenin become self-sustaining in FP B receptor-expressing cells. Our hypothesis is that the capacity of GSK-3␤ to phosphorylate ␤-catenin is overwhelmed by the combination of the increased expression of ␤-catenin (conditioning) and by the agonist-induced shape change and reorganization of ␤-catenin (activation). Although the activation step also results in the increased phosphorylation and inhibition of GSK-3␤, the effects of C3-toxin (Fig. 8) show that agonist-induced Tcf transcriptional activation is actually independent of GSK-3␤ activity and can be mediated solely through a Rho-dependent pathway. This sequence of conditioning and activation increases the concentration of E-cadherin and ␤-catenin in regions of cell-cell adhesion and maintains the aggregated phenotype by generating free cytosolic ␤-catenin at a rate that exceeds the capacity of the cell to degrade it. The resulting increase in Tcf transcriptional activation could then initiate a positive feedback loop; for example, by inducing further expression of E-cadherin or some other adhesion molecule.
Our present findings and conclusions regarding the mechanism of FP B prostanoid receptor signaling have general significance with regards to GPCR signaling. Most importantly it clearly illustrates the potential of a receptor to alter the cellular signaling environment by virtue of its constitutive activity. The fact that receptors may have agonist-independent constitutive activity has been known for some time, and its importance has been recognized, as, for example, in mutations of the luteinizing hormone receptor that are responsible for precocious puberty (12). In this example, however, constitutive agonist-independent activity represents the same activity as that obtained with agonist, which is also the case for virtually all previous descriptions of constitutive activity. The agonist-independent conditioning of FP B -expressing cells that we have described differs importantly in this regard. Thus, it represents an independent signaling pathway, i.e. coupling to PI3K, that is actually disrupted by treatment with agonist. Additionally this constitutive activity profoundly alters the signaling potential of the agonist itself. Thus, PGF 2␣ treatment activates Rho and induces cellular shape change in both FP A -and FP Bexpressing cells, but it only activates Tcf/␤-catenin signaling in FP B -expressing cells. We believe it is likely that other GPCRs will have similar agonist-independent constitutive activities. Furthermore, such activities may be best revealed through the application of gene microarray technology and other approaches that do not rely on conventional assumptions about GPCR signaling. appreciate the efforts of Sambhitab Salvi and Christopher T. Mogan for preparation of C3-toxin.