Modulation of Stat3 Activation by the Cytosolic Phospholipase A2α and Cyclooxygenase-2-controlled Prostaglandin E2 Signaling Pathway*

A variety of human cancers show constitutive activation of signal transducer and activator of transcription-3 (Stat3) and overexpression of cyclooxygenase-2 (COX-2). This study describes a novel cross-talk between the COX-2-controlled prostaglandin E2 (PGE2) and Stat3 signaling pathways that coordinately regulate human cancer cell growth. COX-2-derived PGE2 induces interleukin-6 production through activation of EP4 receptor and subsequent phosphorylation of gp130/Stat3 in human cholangiocarcinoma cells. In parallel, activation of COX-2/PGE2 signaling also enhances Stat3 phosphorylation and reporter activity through EP1 receptor-induced activation of c-Src and EGFR in these cells. Moreover, the observations that EP1 receptor is detected in the nucleus as well as in the Stat3·DNA binding complex and that activation of EP1 receptor in the nuclei enhances Stat3 activation depicts a previously undescribed G protein-coupled receptor in the nucleus for Stat3 activation and tumor cell growth.

Stat3 3 (signal transducer and activator of transcription-3) is one member of the Stat transcription factor family that is normally activated in a regulated fashion when protein ligands bind their specific cell surface receptors and activate tyrosine kinases (1)(2)(3). Similar to other members of the Stat transcription factor family, Stat3 becomes activated by phosphorylation on a single tyrosine residue, dimerizes through reciprocal SH2-phosphotyrosine interaction, and accumulates in the nucleus, where it binds DNA and direct transcription of a wide array of genes. In normal cells the level and duration of Stat3 activation is controlled by mechanisms including dephosphorylation of the receptor complex or nuclear Stat dimers by protein-tyrosine phosphatases (PTPases), interaction of activated Stats with inhibitory molecules from the protein inhibitors of activated Stat (PIAS) family, and feedback inhibition of the JAK/STAT pathway by suppressor of cytokine signaling (SOCS) proteins through inhibition and/or degradation of JAKs. In sharp contrast, Stat3 is constitutively activated in a variety of human cancer cells, including all the major carcinomas as well as some hematologic tumors (2,3), although the mechanisms for persistent Stat3 activation during tumorigenesis are not fully understood.
Whereas laboratory-prepared mutations resulting in constitutively active Stat3 can produce a transforming Stat3 oncogene, no naturally occurring mutations of Stat3 have been described so far (2,3). Recent evidence suggests that constitutive Stat3 activation is frequently a consequence of aberrant autocrine or paracrine stimulation of cytokine and growth factor receptors, such as IL-6-driven JAK/Stat3 signaling in cholangiocarcinoma (4). Other potential mechanisms for persistent Stat3 activation in tumor cells include overexpression or gain of function mutations of receptor-tyrosine kinases that activate Stat3, such as EGFR, or loss of proteins that negatively regulate Stat3, such as PIAS or SOCS.
In addition to Stat3, constitutive overexpression of cyclooxygenase-2 (COX-2) and elevated prostaglandin (PG) production has also been found in a variety of human cancers including cholangiocarcinoma (4 -7). The expression of COX-2 in human cholangiocarcinoma tissue is positively correlated with EGFR (erbB1) and erbB2, two members of the EGFR tyrosine kinase family (8,9). Constitutive expression of erbB2, a receptor-tyrosine kinase of the epidermal growth factor receptor (EGFR) family, in gall bladder and biliary tree epithelia results in elevated COX-2 and development of gall bladder adenocarcinoma and cholangiocarcinoma in mice (10). The expression of COX-2 in cultured cholangiocarcinoma cells is induced by growth factors (8), proinflammatory cytokines (11,12), and bile acid (a tumor promoter in cholangiocarcinogenesis) (11,(13)(14)(15). Further, overexpression of COX-2 or treatment with PGE 2 enhances tumor growth and invasion, in vitro and in SCID mice (8, 12, 16 -20), whereas inhibition of COX-2 reduced cholangiocar-cinoma cell growth (16,17,19,20). These findings indicate an important role of PG signaling in tumorigenesis.
PG biosynthesis is tightly controlled by a series of enzymes including the group IV␣ cytosolic phospholipase A 2 (cPLA 2 ␣) that selectively cleaves arachidonic acid (AA) from membrane phospholipids, and COX-2 that converts AA substrate to PGs. Prostanoids exert their biological actions primarily via their respective G protein-coupled receptors (GPCR) superfamily of seven-transmembrane spanning proteins on the cell surface membrane (21,22). PGE 2 , the most abundant prostaglandin in human carcinoma cells, can potentially interact with four types of receptors (EP 1 , EP 2 , EP 3 , and EP 4 ). The EP 1 receptor is coupled with G q protein and thus signals through phospholipase C and intracellular Ca 2ϩ ; the EP 2 and EP 4 receptors are coupled with G s protein, signaling through elevation of intracellular cAMP level and activation of protein kinase A; the EP 3 receptor is coupled with G i protein and signals through reduction of intracellular cAMP. Recently, EP 1 receptor has also been detected in the nucleus in certain cells (porcine cerebral endothelial cells and murine Swiss 3T3 cells) (23), although the phys- Note the cytoplasmic and nuclear staining of p-Stat3 and the cytoplasmic staining of COX-2 in sequential sections of the same tumor. No stain was seen when the primary antibody was substituted with non-immunized serum (NC, negative control). B, PGE 2 induces Stat3 phosphorylation in SG231 cells. The SG231 cells with 80% confluence were serum-starved for 24 h prior to treatment with 10 M PGE 2 for 5-60 min or 50 ng/ml IL-6 for 30 min. Stat3 phosphorylation was determined by immunoprecipitation with anti-Stat3 antibody and immunoblotting with anti-phosphotyrosine (PY99) antibody (top panel ); Stat3 in the immunoprecipitate was determined by reprobing the same blot with anti-Stat3 antibody (middle panel ). Quantitative analysis of Stat3 phosphorylation was performed by determining the ratio between the Stat3 protein and phosphorylation levels from three different experiments with densitometry (low panel ). PGE 2 increased Stat3 phosphorylation as early as 15 min after treatment (*, p Ͻ 0.01). C, PGE 2 induces Stat3 phosphorylation in CCLP1 cells. The CCLP1 cells with 80% confluence were serum-starved for 24 h prior to PGE 2 treatment (10 M). Stat3 phosphorylation was determined by immunoblotting with anti-phospho-Stat3 (Tyr 705 ) antibody (top panel ); Stat3 was determined by reprobing the same blot with anti-Stat3 antibody (middle panel ). Quantitative analysis of Stat3 phosphorylation was performed by determining the ratio between the Stat3 protein and phosphorylation levels from three different experiments with densitometry (low panel ). PGE 2 increased Stat3 phosphorylation as early as 5 min after treatment (*, p Ͻ 0.01). D, overexpression of COX-2 increases Stat3 phosphorylation. The CCLP1 cells with 80% confluence were transiently transfected with the COX-2 expression plasmid or pcDNA control plasmid. After transfection, the cells were cultured in serum-free medium for 24 h. The cell lysates were then obtained to determine the phosphorylation of Stat3 by immunoblotting with anti-phospho-Stat3 (Tyr 705 ) antibody (top panel ). The same blot was reprobed with antibodies against Stat3 (second panel ), COX-2 (third panel ), and ␤-actin (fourth panel ). Quantitative analysis of Stat3 phosphorylation was performed by determining the ratio between the Stat3 protein and phosphorylation levels from three different experiments with densitometry (lower panel ) (*, p Ͻ 0.05). E, overexpression of cPLA 2 ␣ increases Stat3 phosphorylation. The CCLP1 cells with 80% confluence were transiently transfected with the cPLA 2 ␣ expression plasmid or MT-2 control plasmid. After transfection, the cells were cultured in serum-free medium for 24 h and then cell lysates were obtained to determine the phosphorylation of Stat3 (top panel ). The same blot was reprobed with antibodies against Stat3 (second panel ), cPLA 2 ␣ (third panel ), and ␤-actin (fourth panel ). Quantitative analysis of Stat3 phosphorylation was performed by determining the ratio between the Stat3 protein and phosphorylation levels from three different experiments with densitometry (lower panel ) (*, p Ͻ 0.05).
iological implication of this intriguing phenomenon remains unknown.
Besides the direct tumor-promoting effect, PG signaling has also been implicated in the growth of cholangiocarcinoma cells induced by growth factors including interleukin-6 (IL-6) (4, 18), a classical biliary mitogen that is implicated in cholangiocarcinogenesis (5-7). IL-6 is known to stimulate biliary mitogenesis in an autocrine and paracrine fashion (24 -28). It is produced by periductal hematolymphoid cells during diseases associated with non-neoplastic biliary epithelial cell growth, whereas neoplastic human biliary epithelial cells can acquire the ability to constitutively produce IL-6. Accordingly, patients with cholangiocarcinoma show increased serum levels of IL-6 production (29,30) and in animal models of cholangiocarcinoma IL-6 is detected after inoculation of cholangiocarcinoma cell line into SCID mice (24). IL-6 induces its biological functions through binding to its receptor complex (gp130 and gp80), which triggers the activation of Stat3 and mitogen-activated protein kinase (MAPK) pathways (31)(32)(33)(34). Consistent with the observation that IL-6 increases PGE 2 production via MAPK-induced phosphorylation of cPLA 2 ␣, blocking PGE 2 synthesis inhibits the IL-6-induced cholangiocarcinoma cell growth (18). Thus, the interaction between PGE 2 and IL-6 signaling pathways may play a potential role in cholangiocarcinogenesis, although detailed mechanism for such an interaction remains to be further defined.
This study was designed to dissect the interactions between the cPLA 2 ␣/COX-2/PGE 2 and IL-6/Stat3 signaling pathways in human cholangiocarcinoma cells and examine their role in tumor growth. Our findings indicate that the cPLA 2 ␣/COX-2/ PGE 2 signaling activates Stat3 through two parallel mechanisms, i.e. the EP 4 receptor-mediated induction of IL-6 production (extracellular mechanism) and the EP 1 -receptor-mediated activation of c-Src (intracellular mechanism). Moreover, this study unveils a novel role of EP 1 , a G protein-coupled receptor, in the nucleus for Stat3 activation. Taken together, our results depict a novel cross-talk between the cPLA 2 ␣/COX-2/PGE 2 and IL-6/Stat3 signaling pathways that coordinately regulate tumor cell growth.

EXPERIMENTAL PROCEDURES
Materials-Minimum essential medium ␣ (␣-MEM), Dulbecco's modified minimum essential medium (DMEM), fetal bovine serum, glutamine, antibiotics, the Lipofectamine plus TM reagent and Lipofectamine TM 2000 reagent were purchased from Invitrogen (Carlsbad, CA). Prostaglandin E 2 (PGE 2 ), the cPLA 2 ␣ inhibitors AACOCF 3 and N-{(2S,4R)-4-(biphenyl-2-ylmethyl-isobutyl-amino)-1-[2-(2,4-difluorobenzoyl)-benzoyl]-pyrrolidin-2-ylmethyl}-3-[4-(2,4-dioxothiazolidin-5-ylidenemethyl)-phenyl]acrylamide, HCl, the COX-2 inhibitor NS398, and the Src family tyrosine kinase inhibitor 4-amino-5-(4-chlorophenyl)-7-(t-butyl) pyrazolo [3,4-d]pyrimidine (PP2) were purchased from Calbiochem. The recombinant human IL-6 was purchased from R&D systems (Minneapolis, MN). The EP 1 agonist ONO-DI-004, the EP 1 antagonist ONO-8711 and the EP 3 agonist ONO-AE-248 were provided by the ONO Pharmaceutical Co., Ltd (Osaka, Japan). The EP 2 agonist Butaprost and EP 4 agonist PGE 1 Alcohol were pur-FIGURE 2. cPLA 2 ␣/COX-2/PGE 2 signaling increases Stat3 transcription activity in human cholangiocarcinoma cells. A, PGE 2 increases Stat3 transcription activity in SG231 cells. The SG231 cells were transiently transfected with pStat3-Luc reporter vector. After transfection, the cells were treated with 10 M PGE 2 , 50 ng/ml IL-6, or vehicle in serum-free medium for 24 h, and the cell lysates were obtained to determine the luciferase activity as described under "Experimental Procedures." The data are presented as mean Ϯ S.D. of six independent experiments (*, p Ͻ 0.01). B, overexpression of COX-2 increases Stat3 transcription activity in SG231 cells. The SG231 cells were transiently transfected with the COX-2 expression plasmid or pcDNA control plasmid with co-transfection of pStat3-Luc reporter vector. After transfection the cells were cultured in serum-free medium for 24 h, and then the cell lysates were obtained to determine the luciferase activity. The data are presented as mean Ϯ S.D. of six independent experiments. The cells with COX-2 overexpression confirmed by Western blot (lower panel ) showed significantly increased luciferase reporter activity when compared with the cells transfected with control vector (*, p Ͻ 0.01). C, PGE 2 increases Stat3 transcription activity in CCLP1 cells. The CCLP1 cells were transiently transfected with pStat3-Luc reporter vector. After transfection the cells were treated with 10 M PGE 2 , 50 ng/ml IL-6 or vehicle in serum-free medium for 24 h, and the cell lysates were obtained to determine the luciferase activity. The data are presented as mean Ϯ S.D. of six independent experiments (*, p Ͻ 0.01; **, p Ͻ 0.05). D, overexpression of COX-2 increases Stat3 reporter activity in CCLP1 cells. The CCLP1 cells were transiently transfected with the COX-2 expression plasmid or pcDNA control plasmid with co-transfection of pStat3-Luc reporter vector. After transfection the cells were cultured in serum-free medium for 24 h, and the cell lysates were obtained to determine the luciferase activity. The data are presented as mean Ϯ S.D. of six independent experiments. The cells with COX-2 overexpression confirmed by Western blot (lower panel ) showed significantly increased luciferase reporter activity (*, p Ͻ 0.05). E, overexpression of cPLA 2 ␣ increases Stat3 reporter activity in CCLP1 cells. The CCLP1 cells were transiently transfected with the cPLA 2 ␣ expression plasmid or MT-2 control plasmid with co-transfection of pStat3-Luc reporter vector. After transfection the cells were cultured in serum-free medium for 24 h, and the cell lysates were obtained to determine the luciferase activity. The data are presented as mean Ϯ S.D. of six independent experiments. The cells with cPLA 2 ␣ overexpression confirmed by Western blot (lower panel ) showed significantly increased luciferase reporter activity when compared with the cells transfected with control plasmid (**, p Ͻ 0.01). chased from Cayman Chemical (Ann Arbor, MI). The antibodies against human cPLA 2 ␣, c-Src, anti-phosphotyrosine (PY99), and caveolin-1 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Stat3 antibody and phospho-Stat3 (Tyr 705 ) antibody were purchased from Cell Signaling (Beverly, MA). The antibodies against EP 1 , EP 2 , EP 3 , and EP 4 receptors and COX-2 were purchased from Cayman Chemical. The antibody against human gp130 was purchased from R&D systems. The antibody against ␤-actin was purchased from Sigma. Horseradish peroxidase-linked streptavidin and chemiluminescence detection reagents were purchased from Amersham Biosciences. The human IL-6 immunoassay kit was purchased from R&D Systems.
Cell Culture and Transient Transfection-Human cholangiocarcinoma cell lines including CCLP1 and SG231 were cultured according to our previously described methods (8,17,18,35). Briefly, CCLP1 cells were cultured in DMEM supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and 50 g/ml gentamycin; SG231 cells were cultured in ␣-MEM with 10% fetal bovine serum, 2 mM L-glutamine, 50 g/ml gentamicin, and 10 mM HEPES. The cells were cultured at 37°C in a humidified CO 2 incubator. For transient transfection assays, the cultured cells were transfected with both the cPLA 2 ␣ expression plasmid or MT-2 control plasmid and the COX-2 expression plasmid or pcDNA control plasmid using Lipofectamine plus TM reagent. The cells with optimal overexpression of either cPLA 2 ␣ or COX-2 were confirmed by immunoblotting and subsequently used for further experiments.
Luciferase Reporter Assay-The cultured cells were seeded at a concentration achieving 80% confluence in 12-well plates 18 h before transfection. The cells were transiently transfected with 0.2 g/per well translucent Stat3 (1)-Luc reporter vector purchased from Panomics (Redwood City, CA), which was designed to measure the transcriptional activity of Stat3. After transfection, the cells were treated with specific reagents such as PGE 2 or EP agonists in serum-free medium for 24 h. The cell lysates were then obtained with 1ϫ reporter lysis buffer (Promega). The luciferase activity was assayed in a Berthold AutoLumat LB 953 luminometer (Nashua, NH) using the luciferase assay system from Promega. The relative luciferase activity was calculated after normalization of cellular proteins. All values are expressed as -fold induction relative to basal activity.
Cell Growth Assay-Cell growth was determined using the cell proliferation reagent WST-1, a tetrazolium salt that is cleaved by mitochondrial dehydrogenases in viable cells. Briefly, 100 l of cell suspension (containing 0.5-2 ϫ 10 4 cells) were plated in each well of 96-well plates. After a 24-h culture to allow reattachment, the cells were then treated with specific reagents such as PGE 2 or EP agonists for indicated time points. At the end of each experiment, the cell proliferation reagent WST-1 (10 l) was added to each well, and the cells were incubated at 37°C for 0.5-5 h. A 450 nm was measured using an automatic ELISA plate reader.
Immunoblotting-At the end of each indicated treatment, the cells were scraped off the plates and centrifuged, washed twice with cold phosphate-buffered saline (PBS) containing 0.5 mM phenylmethylsulfonyl fluoride, and 10 g/ml leupeptin and resuspended in 5-fold volume of hypotonic buffer consisting of 50 mM HEPES pH 7.55, 1 mM EDTA, 1 mM dithiothreitol, and protease inhibitor mixture tablets (Roche Applied Science). After sonication, the whole cell lysate was collected by centrifugation at the speed of 15,000 ϫ g at 4°C for 10 min to remove cell debris and stored in aliquots at Ϫ20°C until use. The protein concentrations in the cell extracts were determined by the Bio-Rad protein assay. 30 g of cellular protein was subjected to SDS-PAGE on 4 -20% Tris-glycine gels for cPLA 2 ␣, COX-2, c-Src, Stat3, phospho-Stat3, EP 1 , EP 2 , EP 3 , EP 4 , or on 6% Trisglycine gel for gp130. The separated proteins were electrophoretically transferred onto the nitrocellulose membranes (Bio-Rad). Nonspecific binding was blocked with PBS-T (0.5% Tween 20 in PBS) containing 5% nonfat milk for 1 h at room temperature. The membranes were then incubated overnight at 4°C with individual primary antibodies in PBS-T containing 1% nonfat milk at the dilutions specified by the manufacturers. Following three washes with PBS-T, the membranes were then incubated with the horseradish peroxidase-conjugated secondary antibodies at 1:10,000 dilution in PBS-T containing 1% nonfat milk for 1 h at room temperature. The membranes were then washed three times with PBS-T, and the protein bands were visualized with the ECL Western blotting detection system according to the manufacturer's instructions.
Phosphorylation of gp130-After each indicated treatment, the cell lysates were obtained as described above. Equal amounts of the cell lysates were preincubated with 5 g/ml goat FIGURE 3. PGE 2 induces IL-6 production through the EP 4 receptor in SG231 cells. A, PGE 2 induces IL-6 production in SG231 cells. The SG231 cells with 80% confluence were serum-starved overnight before PGE 2 treatment (10 M) in serum-free medium for 24 h. The IL-6 in the cultured medium was measured by immunoassay. The data are presented as mean Ϯ S.D. of three independent experiments (*, p Ͻ 0.01). B, effect of COX-2 overexpression on IL-6 production in SG231 cells. The SG231 cells were transiently transfected with the COX-2 expression plasmid or pcDNA control plasmid. After transfection the cells were cultured in serumfree medium for 24 h, and the culture media were obtained to measure IL-6 production by immunoassay. The data are presented as mean Ϯ S.D. of three independent experiments. The cells with COX-2 overexpression confirmed by Western blot (lower panel ) showed significantly increased IL-6 production (*, p Ͻ 0.01). C, effect of different EP agonists on IL-6 production in SG231 cells. The SG231 cells were treated with PGE 2 , ONO-DI-004, Butaprost, ONO-AE-248, PGE 1 -OH, or vehicle in serum-free medium for 48 h. The culture medium was then obtained to measure IL-6 production by immunoassay. The data are presented as mean Ϯ S.D. of three independent experiments. The cells treated with PGE 2 or PGE 1 -OH showed significantly increased IL-6 production when compared with control (*, p Ͻ 0.01). A similar effect was also observed at 24 h. D, antisense suppression of EP 4 blocks PGE 2 -induced IL-6 production in SG231 cells. The SG231 cells were transfected with antisense oligonucleotides of EP 1 , EP 2 , EP 3 , or EP 4 . The cells were then treated with 10 M PGE 2 or corresponding agonists (1 M ONO-DI-004 for EP 1 , 5 M Butaprost for EP 2 , 5 M ONO-AE-248 for EP 3 , and 1 M PGE 1 -OH for EP 4 ) in serum-free medium for 48 h. The culture media were obtained to measure IL-6 production by immunoassay. The data are presented as mean Ϯ S.D. of three independent experiments. Antisense suppression of EP 4 blocked both PGE 2 -and PGE 1 -OH-induced IL-6 production (*, p Ͻ 0.01 compared with irrelevant oligonucleotide control cells treated with PGE 2 or PGE 1 -OH). Western blots showed successful depletion of EP 1 , EP 2 , EP 3 , and EP 4 receptors in SG231 cells transfected with the corresponding antisense oligonucleotides. AUGUST 25, 2006 • VOLUME 281 • NUMBER 34

JOURNAL OF BIOLOGICAL CHEMISTRY 24835
anti-human gp130 polyclonal antibody followed by addition of 20 l of protein A/G-agarose. The mixtures were incubated overnight at 4°C. After three washes with the same hypotonic buffer, the pellet was used for immunoblotting using monoclonal anti-phosphotyrosine (PY99) antibody.
Binding of c-Src to Stat3-The binding complexes of Stat3 and c-Src in CCLP1 cells were determined by immunoprecipitation and Western blot. Confluent CCLP1 cells were serumstarved for 24 h, followed by treatment with 1 M ONO-DI-004 or 10 M PGE 2 for 30 min. The cell lysates were subsequently prepared for immunoprecipitation with antibody against c-Src. The immunoprecipitants were then subjected to SDS-PAGE and immunoblotted with anti-Stat3 antibody.
Binding of Stat3 to EP 1 -The binding complexes of Stat3 and EP 1 in CCLP1 cells were determined by immunoprecipitation and Western blot. Confluent CCLP1 cells were serum-starved for 24 h, followed by treatment with 1 M ONO-DI-004 or 10 M PGE 2 for 30 min. The cell lysates were subsequently prepared for immunoprecipitation with antibody against Stat3. The immunoprecipitants were then subjected to SDS-PAGE and immunoblotted with anti-EP 1 antibody.
Preparation of Caveolin-rich Membrane Fractions-Caveolin-enriched membrane fractions were prepared from CCLP1 cells according to the method described previously (37). In brief, CCLP1 cells grown to confluence in 100-mm dishes were washed twice with PBS and homogenized in 2 ml of 500 mM sodium carbonate, pH 11.0 containing inhibitors of proteases and protein phosphatases. The homogenate was then adjusted to 45% sucrose by the addition of 90% sucrose prepared in MBS (25 mM MES, pH 6.5, 0.15 M NaCl) and placed at the bottom of an ultracentrifuge tube. A 5-35% discontinuous sucrose gradient was formed above (4 ml of 5% sucrose/4 ml of 35% sucrose; both in MBS containing 250 mM sodium carbonate) and centrifuged at 39,000 rpm for 20 h in an SW41 rotor (Beckman Instruments). A light-scattering band confined to the 5-35% sucrose interface was collected as caveolinenriched membrane fractions for further experiments.
RNA Interference-Stat3 siRNA (siRNA ID: 42861) and gp130 siRNA were purchased from Ambion (Austin, TX). The targeted sequences that effectively mediate the silencing of gp130 expression are the combination of 5Ј-GCAAGUGGGAU-CACCUAUGTT-3Ј (sense sequence) (siRNA ID: 111101) and 5Ј-GGCAUGCCUAAAAGUUACUTT-3Ј (sense sequence) (siRNA ID: 106711). Cells with 50% confluence were transfected with either Stat3 siRNA or gp130 siRNA, or a 21-nucleotide irrelevant RNA duplex as a control using Lipofectamine TM 2000. Depletion of either Stat3 or gp130 was confirmed by immunoblotting and subsequently used for further experiments.
Biotinylated Oligonucleotide Precipitation Assays-These experiments were performed as described previously with minor modifications (40). The sequences of biotinylated oligo-FIGURE 4. PGE 2 /EP 4 signaling phosphorylates and activates gp130 in SG231 cells. A, PGE 2 induces gp130 phosphorylation in SG231 cells. The SG231 cells were serum-starved for 24 h prior to PGE 2 (10 M for 5-60 min) or IL-6 (50 ng/ml for 30 min) treatment. gp130 phosphorylation was determined by immunoprecipitation with anti-gp130 antibody and immunoblotting with anti-phosphotyrosine (PY99) antibody (top panel ). gp130 in the immunoprecipitate was determined by reprobing the same blot with anti-gp130 antibody (middle panel ). Quantitative analysis of gp130 phosphorylation was performed by determining the ratio between the gp130 protein and phosphorylation levels from three different experiments with densitometry (bottom panel ). PGE 2 increased gp130 phosphorylation as early as 5 min after treatment (*, p Ͻ 0.01). B, COX-2 overexpression increases gp130 phosphorylation in SG231 cells. The SG231 cells were transiently transfected with the COX-2 expression plasmid or pcDNA control plasmid. After transfection the cells were cultured in serum-free medium for 24 h, and then cell lysates were obtained to determine the phosphorylation of gp130 and Stat3. gp130 phosphorylation was determined by immunoprecipitation with anti-gp130 antibody and immunoblotting with anti-phosphotyrosine (PY99) antibody (top row); gp130 in the immunoprecipitate was determined by reprobing the same blot with anti-gp130 antibody (second row). Equal amounts of the same cell lysates were used for regular Western blot to detect Stat3 phosphorylation with anti-phospho-Stat3 (Tyr 705 ) antibody (third row); the same blot was reprobed with antibodies against Stat3 (fourth row), COX-2 (fifth row), and ␤-actin (bottom row). C, antisense suppression of EP 4 blocked PGE 2 -induced phosphorylation of gp130 and Stat3 in SG231 cells. The SG231 cells were transfected with EP 4 antisense oligonucleotides or irrelevant oligonucleotides prior to PGE 2 or vehicle treatment in serum-free medium for 24 h. After transfection the cell lysates were obtained to determine the phosphorylation of gp130 and Stat3. gp130 phosphorylation was determined by immunoprecipitation with anti-gp130 antibody and immunoblotting with anti-phosphotyrosine (PY99) antibody (top row); gp130 in the immunoprecipitate was determined by reprobing the same blot with anti-gp130 antibody (second row). Equal amounts of the same cell lysates were used for regular Western blot to detect Stat3 phosphorylation with anti-phospho-Stat3 (Tyr 705 ) antibody (third row); the same blot was reprobed with antibodies against Stat3 (fourth row), EP 4 (fifth  reactions in nuclear extracts or cell extracts were carried out at 4°C for 16 h with 1 g of Stat3 biotinylated double-strand oligonucleotides and 10 g of poly(dl-dC)⅐poly(dl-dC). The DNAbound proteins were precipitated using ImmunoPure streptavidin-agarose beads (Pierce) at 4°C for 1 h and subjected to detection of EGFR, EP 1 , or Stat3 by Western blotting.
Immunohistochemical Analysis for p-Stat3 and COX-2-Twelve archival formalin-fixed, paraffin-embedded specimens of human cholangiocarcinoma and surrounding nontumor liver tissue were obtained from the University of Pittsburgh Medical Center. The tissue specimens were utilized for immunohistochemical analysis for p-Stat3 and COX-2 according to the protocol approved by the University of Pittsburgh (IRB no.: 0506052). None of the cases used in this study had patient identifiers and strict confidentiality was maintained in accordance with the approval granted by the Institutional Review Board. 5-m-thick tissue sections of formalin-fixed and paraffin-embedded sections were deparaffinized and rehydrated, followed by microwave retrieval of antigen according to standard procedures. For immunostaining with p-Stat3, the slides were incubated at room temperature for 1 h with 1:600 diluted rabbit anti-human p-Stat3 (obtained from Cell Signaling catalog no. 9131) in labeling buffer. Following repeated washings, the slides were incubated with CSA Rabbit Link followed by addition of streptavidin/biotin (Dako CSA system for immunohistochemistry) according to the manufacturer's instruction. AEC substrate was used for color development followed by counterstaining with hematoxylin. The immunostain for COX-2 was performed according to our previously described method (16,41), using 3,3Ј-diaminobenzidine (DAB) substrate for color development.

RESULTS
We first utilized immunohistochemical stains to determine the expression and localization of p-Stat3 and COX-2 in human cholangiocarcinoma tissues and nonneoplastic bile ducts. Twelve paired cholangiocarcinoma and their matched non-tumor liver tissues were analyzed. Immunoreactivity for p-Stat3 and COX-2 were detected in the tumor cells from all the twelve patients. Fig. 1A highlights the expression of p-Stat3 and COX-2 in sequential sections of human cholangiocarcinoma tissue from the same patient. Although the level of p-Stat3 is not increased in human cholangiocarcinoma cells when compared with the nonneoplastic bile duct epithelial cells, the expression pattern of p-Stat3 in cholangiocarcinoma cells is different from the nonneoplastic biliary epithelial cells. In nonneoplastic biliary epithelial cells, p-Stat3 is expressed exclusively in the nuclei, whereas in cholangiocarcinoma cells, p-Stat3 is expressed in both nuclei and cytoplasm.
The coexpression of COX-2 and p-Stat3 in human cholangiocarcinoma cells suggests a possible link between these two signaling pathways for tumor growth. The direct effect of PGE 2 on Stat3 phosphorylation was next examined in two human cholangiocarcinoma cell lines, SG231 and CCLP1. As shown in Fig. 1, B and C, PGE 2 treatment increased the phosphorylation of Stat3 in both cell lines. The PGE 2 -induced Stat3 phosphorylation was observed 5 min after treatment in CCLP1 cells and 15 min after treatment in SG231 cells; the effect sustained at 60 min in both cell types. The magnitude of PGE 2 -induced Stat3 phosphorylation was ϳ70% of that induced by IL-6, a cytokine that signals exclusively through activation of Stat3. Given that the synthesis of PGE 2 in human cholangiocarcinoma cells is tightly controlled by coupled activation of cPLA 2 ␣ and COX-2, further experiments were performed to determine whether overexpression of cPLA 2 ␣ and COX-2 would also affect Stat3 phosphorylation. As shown in Fig. 1, D and E, increased Stat3 phosphorylation was also observed in CCLP1 cells with overexpression of either cPLA 2 ␣ or COX-2. Both cPLA 2 ␣-transfected cells and COX-2-transfected cells exhibited increased PGE 2 production compared with the control vector cells (598 pg/ml in cPLA 2 ␣-transfected cells versus 310 pg/ml in corresponding control vector cells; 430 pg/ml in COX-2-transfected cells versus 295 pg/ml in corresponding vector control cells) (SG231 cell was not utilized for overexpression experiments because of its high basal level of cPLA 2 ␣ expression).
The potential effect of PGE 2 signaling on Stat3 transcription activity was next examined in human cholangiocarcinoma cells transfected with a luciferase reporter construct under the control of activated Stat3. As shown in Fig. 2, treatment of SG231 and CCLP1 cells with PGE 2 resulted in ϳ2.6and 2.2-fold increase of Stat3 transcription activity, respectively. Similarly, overexpression of COX-2 and cPLA 2 ␣ in these cells also induced 2-3-fold increase of Stat3 reporter activity. These findings demonstrate that the cPLA 2 ␣ and COX-2-controlled PGE 2 signaling activates Stat3 in human cholangiocarcinoma cells.
Further experiments were designed to determine the mechanisms by which PGE 2 activates Stat3 in human cholangiocarcinoma cells. Given that IL-6-mediated Stat3 phosphorylation is critically involved in biliary epithelial cell growth and cholangiocarcinogenesis, we investigated whether PGE 2 might activate Stat3 via IL-6, an endogenous ligand that induces Stat3 phosphorylation. Indeed, as shown in Fig. 3A, treatment of SG231 cells with PGE 2 for 24 h induced a 5.5-fold increase of IL-6 production, and the effect is dose-dependent. Similarly, overexpression of COX-2 also significantly increased IL-6 production in these cells (Fig. 3B). These findings indicate that COX-2 and PGE 2 induces IL-6 production in human cholangiocarcinoma cells.
Given that four types of PGE 2 receptors (EP 1 , EP 2 , EP 3 , and EP 4 ) are expressed in SG231 cells, additional experiments were performed to investigate the role of each individual EP receptor subtype in PGE 2 -induced IL-6 production in these cells. Fig. 3C shows that the EP 4 receptor agonist, PGE 1 -OH, increased IL-6 production in SG231 cells, whereas the agonists for EP 1 (ONO-DI-004), EP 2 (Butaprost) and EP 3 (ONO-AE-248) exhibited no significant effect. Consistent with these findings, the PGE 2 -induced IL-6 production in SG231 cells was blocked by antisense inhibition of EP 4 , but not by antisense inhibition of EP 1 , EP 2 , or EP 3 (Fig. 3D). These findings suggest the involvement of EP 4 , but not EP 1 , EP 2 , or EP 3 , in PGE 2 -induced IL-6 production in SG231 cells.
Because IL-6 mediates Stat3 phosphorylation through activation of its receptor, gp130, we next examined the effect of PGE 2 signaling on gp130 phosphorylation. As shown in Fig. 4A, treatment of SG231 cells with PGE 2 induced a rapid phosphorylation of gp130; the effect occurred within 5 min and sustained at 60 min. Similarly, overexpression of COX-2 also increased the phosphorylation of gp130, in addition to Stat3 (Fig. 4B). Consistent with the documented role of EP 4 receptor in IL-6 production (Fig. 3), anti- AUGUST 25, 2006 • VOLUME 281 • NUMBER 34

JOURNAL OF BIOLOGICAL CHEMISTRY 24839
sense depletion of EP 4 receptor also partially prevented the PGE 2induced phosphorylation of gp130 and Stat3 (Fig. 4C). The contribution of gp130 to PGE 2 -induced Stat3 phosphorylation was further documented using specific siRNA for gp130. As shown in Fig. 4D, siRNA inhibition of gp130 partially prevented the PGE 2 -induced Stat3 phosphorylation and transcription activity. Taken together, these results suggest that the EP 4 receptormediated IL-6 production and gp130 phosphorylation is an important mechanism by which PGE 2 activates Stat3 in SG231 cells. It is of further interest that the IL-6-induced Stat3 activation was blocked by the cPLA 2 ␣ inhi-bitors, AACOCF 3 and the 1,2,4tri-substituted pyrrolidine derivative (C 49 H 44 F 2 N 4 O 5 S), and by the COX-2 inhibitor, NS-398 (Fig. 4E), suggesting that PG signaling also modulate the constitutive IL-6/gp130/Stat3 signaling in SG231 cells.
However, the IL-6-induced gp130 signaling does not appear to represent a key mechanism for PGE 2 -induced Stat3 activation in the CCLP1 cells, because: (1) these cells express low levels of IL-6 and gp130; and (2) PGE 2 failed to induce IL-6 production and gp130 phosphorylation in these cells (data not shown). Given that PGE 2 treatment induced a similar increase of Stat3 phosphorylation in both SG231 and CCLP1 cells (Figs. 1 and 2), mechanism(s) independent of IL-6 production and gp130 phosphorylation likely exists, at least in the CCLP1 cells.
Further experiments were performed to investigate the mechanisms for PGE 2 -induced Stat3 activation in CCLP1 cells. As shown in Fig. 5A, treatment of the CCLP1 cells with PGE 2 or the EP 1 receptor agonist ONO-DI-004 increased the Stat3 reporter activity, whereas the EP 2 agonist Butaprost, EP 3 agonist ONO-AE-248, and EP 4 agonist PGE 1 -OH exhibited no significant effect. The involvement of EP 1 receptor in the FIGURE 6. PGE 2 and EP 1 agonist ONO-DI-004 induces the binding of EGFR and EP 1 to Stat3-responsive element. A, treatment of intact CCLP1 cells with ONO-DI-004 and PGE 2 induces the binding of EGFR or EP 1 to Stat3-responsive element. The CCLP1 cells serum-starved for 24 h were treated with 1 M ONO-DI-004 or 10 M PGE 2 for 30 min. The cell lysates were obtained and precipitated with biotinylated Stat3 oligonucleotides followed by immunoblotting for EGFR, EP 1 or Stat3 (upper panels). Equal amounts of the cell lysates were used to detect EGFR, EP 1 or Stat3 by direct Western blot (lower panels). B, dominant negative inhibition of Stat3 prevents ONO-DI-004-and PGE 2 -induced binding of EGFR or EP 1 to Stat3-responsive element. CCLP1 cells infected with AdGFP or AdSTAT3-DN were treated with 1 M ONO-DI-004 or 10 M PGE 2 for 30 min. The cell lysates were precipitated with biotinylated Stat3 oligonucleotides followed by immunoblotting for EGFR or EP 1 (upper panels). Equal amounts of cell lysates were used to detect EGFR or EP 1 by direct Western (lower panels). C, confocal immunofluorescent detection of EP 1 receptor in CCLP1 cells. The cells were fixed in 2 percent paraformaldehyde and permeabilized with 0.1 percent Triton X-100 prior to incubation with specific antibodies. The cells were stained for EP 1 (green) and F-actin (red), and their nuclei were stained with DAPI (blue). D, a single green color confocal immunofluorescent image showing the presence of EP 1 receptor in the nuclei. E, treatment of the nuclei isolated from CCLP1 cells with ONO-DI-004 and PGE 2 induces the binding of EGFR and EP 1 to Stat3-responsive element. Nuclei isolated from CCLP1 cells were incubated with 10 M ONO-DI-004 or 50 M PGE 2 for 30 min on ice. The nuclear extracts were obtained and precipitated with biotinylated Stat3 oligonucleotides followed by immunoblotting for EGFR, EP 1 or Stat3 (upper panels). Equal amounts of the nuclear extracts were used to determine PARP level by direct Western blot (lower panel ). F, co-localization of EP 1 , EGFR and Src in caveolin-enriched membrane fractions. The caveolin-rich light membrane fractions from CCLP1 cells were prepared by sucrose density gradient centrifugation as described under "Experimental Procedures." The proteins from non-caveolin-enriched membrane fractions (1, 2, and 3) or caveolin-enriched membrane fractions (4, 5, and 6) were applied to immunoblot to detect EGFR, EP 1 , Src, and caveolin-1. G, colocalization of EGFR and EP 1 and Src in nuclear fractions of CCLP1 cells. The proteins from caveolin-enriched membrane fractions (1, 2, and 3) and from nuclear fractions (4, 5, and 6) were applied to the immunoblot to detect EGFR, EP 1 , Src, and Stat3. FIGURE 7. EP 1 /Src/Stat3 pathway is also present in SG231 cells. A, c-Src inhibitor, PP2, blocked PGE 2 -induced Stat3 reporter activity in SG231 cells. The SG231 cells were transiently transfected with pStat3-Luc reporter vector. After transfection the cells were treated with PGE 2 or vehicle in the absence or presence of PP2 in serum-free medium for 24 h, and the cell lysates were obtained to determine the luciferase activity. The data are presented as mean Ϯ S.D. of six independent experiments (*, p Ͻ 0.01). B, antisense suppression of EP 1 or EP 4 blocks PGE 2 -induced Stat3 reporter activity in SG231 cells. The SG231 cells were transfected with antisense oligonucleotides of EP 1 , EP 2 , EP 3 , or EP 4 with co-transfection of pStat3-Luc reporter vector. After transfection the cells were treated with PGE 2 or vehicle in serum-free medium for 24 h, and the cell lysates were obtained to determine the luciferase activity. The data are presented as mean Ϯ S.D. of six independent experiments (*, p Ͻ 0.01). Western blots showed successful depletion of EP 1 , EP 2 , EP 3 , and EP 4 receptors in SG231 cells transfected with the corresponding antisense oligonucleotides. CCLP1 cells is further supported by the observation that the Stat3 reporter activity induced by PGE 2 or ONO-DI-004 was blocked by the selective EP 1 receptor antagonist ONO-8711 (Fig. 5B). Consistent with these findings, antisense depletion of EP 1 receptor also inhibited Stat3 phosphorylation (Fig. 5C). Given that PGE 2 is known to activate Src in human cholangiocarcinoma cells (8), we next examined whether Src might play a role in Stat3 activation in CCLP1 cells. As shown in Fig. 5D, although no Src and Stat3 binding complex was detected under baseline culture conditions, treatment of CCLP1 cells with PGE 2 or the EP 1 agonist, ONO-DI-004, induced the association of Src to Stat3. Similarly, PGE 2 or ONO-DI-004 treatment also induced the formation of EP 1 and Stat3 binding complex (Fig.  5E). In addition, the PGE 2 -induced Stat3 phosphorylation in CCLP1 cells was partially blocked by the Src inhibitor, PP2, and by the EP 1 antagonist, ONO-8711, in vitro (Fig. 5F). These findings suggest that activation of Src by EP 1 is an important mechanism by which PGE 2 signaling activates Stat3 and promotes growth in CCLP1 cells.
The role of EP 1 for Stat3 activation in CCLP1 cells is further supported by the data from biotinylated oligonucleotide precipitation assays. As shown in Fig. 6A, treatment of CCLP1 cells with PGE 2 or the EP 1 agonist, ONO-DI-004, induced the binding of Stat3 to its specific consensus oligonucleotide (no binding was detected in the cells treated with Me 2 SO vehicle). Given that EP 1 has recently been shown to transactivate EGFR (8) and that EGFR is known to activate Stat3 (through direct phosphorylation of Stat3 by EGFR along the plasma membrane or direct interaction between EGFR and Stat3 in the nucleus) (42), we next investigated whether EGFR might also be present in the PGE 2 /EP 1 -induced Stat3⅐DNA binding complex. Indeed, treatment with PGE 2 or ONO-DI-004 also induced the binding of EGFR to the Stat3⅐DNA complex. It is of further interest that the EP 1 receptor is also present in the PGE 2 /EP 1 -induced Stat3⅐DNA binding complex. The association of EGFR and EP 1 to the Stat3 binding complex appears dependent on Stat3, given that dominant negative inhibition of Stat3 completely prevented their binding ability (Fig. 6B). These findings suggest that EP 1 receptor and EGFR may physically interact with Stat3 in the nucleus, thus modulating PGE 2 -induced Stat3 transcription activity.
Confocal immunofluorescence microscopy was utilized to determine the distribution of EP 1 receptor in CCLP1 cells. EP 1 receptor is present in the nucleus as well as in the cytoplasm and plasma membrane (Fig. 6, C and D). To further document whether the nuclear EP 1 receptor is involved in Stat3 activation, intact nuclei were isolated from CCLP cells and the obtained nuclei were incubated with PGE 2 or ONO-DI-004 to determine the binding of Stat3 to its consensus DNA site. As shown in Fig. 6E, treatment of isolated nuclei induced the formation of Stat3⅐DNA binding complex, which also contains EGFR and EP 1 . These findings provide further evidence for the role of nuclear EP 1 receptor in COX-2/PGE 2 -induced Stat3 activation. Because caveolae membrane system has been implicated in the subcellular redistribution of transmembrane proteins, we sought to investigate whether this system is involved in the redistribution of EP 1 and EGFR in CCLP1 cells. As shown in Fig. 6F, EP 1 , EGFR, and c-Src are present in the caveolinenriched membrane fractions, but not in the non-caveolin-enriched membrane fractions. Furthermore, caveolin-1 and associated proteins, including EP 1 , EGFR, and c-Src, are also present in the purified nuclei (Fig. 6G). These findings suggest a potential role of caveolae in the transportation of EP 1 and EGFR to the nucleus for Stat3 activation.
We next examined whether the intracellular Src-dependent Stat3 pathway is also present in the SG231 cells. As shown in Fig. 7, the PGE 2 -induced Stat3 reporter activity was inhibited by the Src inhibitor, PP2, as well as by antisense inhibition of the EP 1 receptor. These findings suggest that activation of Src by EP 1 is also involved in PGE 2 -induced Stat3 activation in SG231 cells. In addi- tion, the PGE 2 -induced Stat3 reporter activity in the SG231 cells was also partially inhibited by antisense depletion of EP 4 receptor, which is consistent with the involvement of EP 4 in IL-6 production and gp130 phosphorylation (Figs. 3 and 4).
The direct contribution of gp130 and Stat3 in PGE 2 -induced cholangiocarcinoma cell growth was further investigated by using specific siRNAs. In SG231 cells, the PGE 2 -induced cell growth was inhibited by siRNA depletion of gp130 or Stat3 (Fig.   8A) as well as by antisense depletion of EP 1 or EP 4 (Fig. 9A); these results further support the involvement of EP 1 /Stat3 and EP 4 /gp130 signaling in PGE 2 -induced growth of SG231 cells. In CCLP1 cells, siRNA inhibition of Stat3 blocked the PGE 2 -induced cell growth (Fig. 8B) (gp130 siRNA was not used in these cells because of their low level of gp130 expression). Antisense depletion of EP 1 receptor in CCLP1 cells also inhibited the PGE 2 -induced cell growth (Fig. 9B). These data further support the role of EP 1 /Stat3 signaling in PGE 2 -induced growth of CCLP1 cells.

DISCUSSION
Prostaglandin and Stat3 signaling pathways have been implicated in the growth of several human cancers including cholangiocarcinoma (4 -7), although the mechanisms for their actions remain to be further understood. Given that both signaling pathways are up-regulated during carcinogenesis, we hypothesized that cross-interaction of these pathways may play an important role in the development and progression of human cancers. Our data presented in this study indicate that the cPLA 2 ␣/COX-2-controlled PGE 2 signaling activates Stat3 through EP 4 receptor-mediated extracellular release of IL-6 as well as through EP 1 receptor-mediated intracellular activation of c-Src in human cholangiocarcinoma cells. Moreover, our findings provide novel evidence for the involvement of EP 1 receptor, a G protein-coupled receptor, in the nucleus for Stat3 activation. The importance of Stat3 in PGE 2 -induced tumor cell growth is further highlighted by the observation that siRNA inhibition of Stat3 significantly inhibited the PGE 2 -induced tumor growth. These findings, along with the results described in our previous study that IL-6 increases PG synthesis through phosphorylation of cPLA 2 ␣ in these cells (18), depict a cross-talk between the autocrine/paracrine loops of the IL-6/Stat3 and cPLA 2 ␣/COX-2/ PGE 2 signaling pathways that coordinately regulate human cancer cell growth (Fig. 10).
Cholangiocarcinoma is a highly malignant epithelial neoplasm arising within the biliary tract (4 -7). It comprises ϳ10 - FIGURE 9. The effect of antisense inhibition of EP receptor subtypes on PGE 2 -induced cell growth. A, antisense inhibition of EP receptors in SG231 cells. The SG231 cells cultured in 96-well plate were transfected with antisense oligonucleotides of EP 1 , EP 2 , EP 3 , or EP 4. After transfection the cells were treated with PGE 2 or vehicle in serum-free medium for 48 h, and the cell growth was determined using WST-1 assay. The data are presented as mean Ϯ S.D. of six independent experiments. Antisense suppression of either EP 1 or EP 4 blocked PGE 2 -induced SG231 cell growth (*, p Ͻ 0.01). Western blots showed successful depletion of EP 1 , EP 2 , EP 3 , and EP 4 receptors in SG231 cells transfected with the corresponding antisense oligonucleotides. B, antisense inhibition of EP receptors in CCLP1 cells. The CCLP1 cells cultured in 96-well plate were transfected with antisense oligonucleotides of EP 1 , EP 2 , EP 3 , or EP 4 . After transfection, the cells were treated with PGE 2 or vehicle in serum-free medium for 48 h, and the cell growth was determined using WST-1 assay. The data are presented as mean Ϯ S.D. of six independent experiments. Antisense inhibition of EP 1 or EP 4 inhibited PGE 2 -induced CCLP1 cell growth (*, p Ͻ 0.01). Western blots showed the successful depletion of EP 1 , EP 3 , and EP 4 receptors in CCLP1 cells transfected with the corresponding antisense oligonucleotides.
15% of hepatobiliary neoplasms and its incidence and mortality is rising (5,43). The tumor often develops from background conditions that cause long-standing inflammation, injury and reparative biliary epithelial cell proliferation, such as primary sclerosing cholangitis (PSC), clonorchiasis, hepatolithiasis, or complicated fibropolycystic diseases. Chronic inflammation and cellular injury within bile ducts, together with partial obstruction of bile flow, appear to be important predisposing factors in the pathogenesis of cholangiocarcinoma. IL-6 is a classical biliary mitogen that has been shown to increase the growth and survival of biliary epithelial cells and cholangiocarcinoma cells. It mediates actions through binding to gp130 receptor, resulting in phosphorylation and activation of Stat3. In this study, we showed that COX-2 overexpression or PGE 2 treatment increased IL-6 production, suggesting that the PGE 2induced cell growth is, at least in part, mediated by IL-6. This assertion is supported by the observations that COX-2 and PGE 2 also induced gp130 phosphorylation as well as Stat3 phosphorylation and reporter activity in SG231 cells. Although all four EP receptors (EP 1 , EP 2 , EP 3 , and EP 4 ) are expressed in the SG231 cells, our data indicate that the COX-2 and PGE 2 -induced IL-6 production and gp130 phosphorylation is predominantly mediated by EP 4 . The later conclusion is based on the following evidence: 1) the EP 4 receptor agonist PGE 1 -OH increased IL-6 production and gp130 phosphorylation; 2) the PGE 2 -induced IL-6 production and gp130/Stat3 phosphorylation was blocked by antisense suppression of EP 4 receptor but not by antisense suppression of EP 1 , EP 2 , or EP 3 in SG231 cells; and 3) siRNA inhibition of gp130 partially prevented the PGE 2induced Stat3 phosphorylation and transcription activity. In this regard, it is noteworthy that our results are consistent with the findings reported in a previous study that disruption of EP 4 (but not EP 1 , EP 2 , or EP 3 ) in mice results in reduced circulating levels of IL-6 and decreased IL-6 production by liver and macrophages (44).
We show that COX-2 and PGE 2 also induces Stat3 phosphorylation in human cholangiocarcinoma cells through an intracellular mechanism that involves EP 1 receptor-mediated activation of c-Src and is independent of IL-6/gp130. This assertion is supported by the following evidence: 1) overexpression of COX-2 or treatment with PGE 2 or the selective EP 1 receptor agonist, ONO-DI-004, induced Stat3 phosphorylation/activation, but not IL-6 production or gp130 phosphorylation in CCLP1 cells; 2) the PGE 2 -induced Stat3 phosphorylation and reporter activity was blocked by the EP 1 receptor antagonist, ONO-8711, and by antisense inhibition of EP 1 receptor; 3) PGE 2 or the selective EP 1 agonist, ONO-DI-004 induced the formation of Src⅐Stat3 and EP 1 ⅐Stat3 binding complex; 4) PGE 2 or ONO-DI-004 induced Stat3 DNA binding that was blocked by dominant negative inhibition of Stat3; 5) the Src inhibitor, PP2, blocked PGE 2 -induced Stat3 phosphorylation and reporter activity; and 6) the EP 1 receptor antagonist ONO-8711 inhibited Stat3 phosphorylation and cell growth in CCLP1 cells. These results establish, for the first time, that EP 1 receptor activates Stat3 via Src. In this context, it is of interest to note that another GPCR, angiotensin II receptor, has been shown to phosphorylate Stat3 in vascular smooth muscle cells (45,46).
The observations that PGE 2 or the EP 1 receptor agonist, ONO-DI-004, induces the formation of Src⅐Stat3 and EP 1 ⅐Stat3 binding complexes provide the most direct evidence for the involvement of EP 1 receptor in Src activation. However, the mechanisms by which EP 1 receptor or other G proteins activate Src remain largely unknown. c-Src is a member of a family of cytoplasmic tyrosine kinases that have several domains including a catalytic domain, a regulatory domain, and SH2 and SH3 binding domains (47). Activated c-Src is capable of interacting with and activating several substrates including Stat3 (48 -50). Recent evidence suggests that G proteins are able to directly bind Src, leading to its activation (51), although the exact domains mediating such an interaction have not been identified. It is conceivable that this process may also involve other unknown proteins. Indeed, our data indicate that EGFR is present in PGE 2 /EP 1 agonist-induced Stat3⅐DNA binding complex and this effect is likely mediated by the direct protein interactions between EGFR and Stat3 (EGFR failed to bind Stat3 oligonucleotide). The latter results, along with the findings that EP 1 receptor transactivates EGFR in human cholangiocarcinoma cells (8) and that EGFR activates Stat3 in other cells (52,53) suggest the involvement of EGFR in PGE 2 /EP 1 -mediated Stat3 activation.
In addition to their expression in the plasma membrane, several GPCRs are recently found in the nuclei, including angiotensin II receptor type 1 (AT 1 ) (54), endothelin receptor subtype B (ET B R) (55), lysophosphatidic acid receptor FIGURE 10. Cross-talk between the IL-6/Stat3 and cPLA 2 ␣/COX-2/PGE 2 signaling pathways in cholangiocarcinoma growth. PGE 2 synthesis is controlled by coupled activation of cPLA 2 ␣ and COX-2 along cell membrane. The produced PGE 2 is released into the extracellular space, which then binds to the membrane G protein-coupled EP receptors on the same cell (autocrine) or on a neighboring cell (paracrine). There are four types of PGE 2 receptors (EP 1 , EP 2 , EP 3 , EP 4 ) in human cholangiocarcinoma cells. Our data indicate that the cPLA 2 ␣/COX-2/PGE 2 signaling activates Stat3 through two parallel mechanisms in human cholangiocarcinoma cells: 1) the EP 1 receptor-mediated activation of c-Src and EGFR; and 2) the EP 4 receptor-mediated production of IL-6, which subsequently binds and activates gp130. On the other hand, IL-6 can further enhance PGE 2 signaling through MAP kinase-mediated cPLA 2 ␣ phosphorylation, which provides arachidonic acid substrate for COX-2 to produce PGE 2 . The depicted autocrine/paracrine loop of the IL-6/Stat3 and cPLA 2 ␣/ COX-2/PGE 2 signaling pathways likely play an important role in cholangiocarcinoma growth. type 1 (LPA 1 R) (56), and PGE 2 receptors (EP 1 , EP 3 , EP 4 ) (23,57). However, the potential biological function of nuclear G protein-coupled receptors is currently unknown. In this study, we provide novel evidence for the functional role of EP 1 receptor in the nucleus for Stat3 activation. This finding is noteworthy, given the documented nuclear localization of the key eicosanoid-forming enzymes, cPLA 2 ␣ and COX-2 (58). Therefore, it is conceivable that PGE 2 generated by coupled activation of cPLA 2 ␣ and COX-2 in the nucleus may be sufficient for local activation of nuclear EP 1 receptor that can in turn modulate gene transcription. The observation that EP 1 and EGFR are present in the PGE 2 /EP 1 agonistinduced Stat3⅐DNA binding complex suggests that these receptors upon activation can modulate Stat3 transcription activity in the nuclei. The latter finding is consistent with the recently reported nuclear interaction of EGFR and Stat3 for the transcriptional activation of inducible nitric-oxide synthase (iNOS) (42).
In summary, the results presented in this article demonstrate a novel cross-talk between the cPLA 2 ␣/COX-2/PGE 2 and IL-6/ Stat3 signaling pathways that is importantly involved in the control of human cancer cell growth. These findings are expected to provide important implications for developing future combinational therapy simultaneously blocking COX-2/ PGE 2 and Stat pathways, which is expected to provide synergistic antitumor effect with lesser side effect than inhibiting each individual pathway alone.