Cyclooxygenase-2-derived Prostaglandin E2 Promotes Human Cholangiocarcinoma Cell Growth and Invasion through EP1 Receptor-mediated Activation of the Epidermal Growth Factor Receptor and Akt*

Cyclooxygenase-2 (COX-2)-mediated prostaglandin synthesis has recently been implicated in human cholangiocarcinogenesis. This study was designed to examine the mechanisms by which COX-2-derived prostaglandin E2 (PGE2) regulates cholangiocarcinoma cell growth and invasion. Immunohistochemical analysis revealed elevated expression of COX-2 and the epidermal growth factor (EGF) receptor (EGFR) in human cholangiocarcinoma tissues. Overexpression of COX-2 in a human cholangiocarcinoma cell line (CCLP1) increased tumor cell growth and invasion in vitro and in severe combined immunodeficient mice. Overexpression of COX-2 or treatment with PGE2 or the EP1 receptor agonist ONO-DI-004 induced phosphorylation of EGFR and enhanced tumor cell proliferation and invasion, which were inhibited by the EP1 receptor small interfering RNA or antagonist ONO-8711. Treatment of CCLP1 cells with PGE2 or ONO-DI-004 enhanced binding of EGFR to the EP1 receptor and c-Src. Furthermore, PGE2 or ONO-DI-004 treatment also increased Akt phosphorylation, which was blocked by the EGFR tyrosine kinase inhibitors AG 1478 and PD 153035. These findings reveal that the EP1 receptor transactivated EGFR, thus activating Akt. On the other hand, activation of EGFR by its cognate ligand (EGF) increased COX-2 expression and PGE2 production, whereas blocking PGE2 synthesis or the EP1 receptor inhibited EGF-induced EGFR phosphorylation. This study reveals a novel cross-talk between the EP1 receptor and EGFR signaling that synergistically promotes cancer cell growth and invasion.

treatment for patients with the advanced disease or any effective therapy for its chemoprevention. Although it is well known that the chronic inflammatory conditions involving the bile ducts predispose patients to the development of cholangiocarcinoma, the molecular mechanisms linking chronic inflammation to malignant transformation remain to be further defined.
Whereas evidence for COX-2 and prostaglandin signaling in cholangiocarcinogenesis is compelling, the mechanism for their actions remains largely unknown. Prostanoids exert their biological actions primarily via their respective G-protein-coupled receptor (GPCR) superfamily of seven-transmembrane spanning G-proteins on the cell-surface membrane (19,20). The most abundant prostaglandin in cholangiocarcinoma cells is PGE 2 (15). There are four EP receptor subtypes that can bind to PGE 2 : EP 1 , EP 2 , EP 3 , and EP 4 . The EP 1 receptor is coupled with the G q protein and thus signals through phospholipase C and intracellular Ca 2ϩ . The EP 2 and EP 4 receptors are coupled with the G s protein, signaling through elevation of intracellular cAMP levels and activation of protein kinase A. The EP 3 receptor is coupled with the G i protein and signals through reduction of intracellular cAMP levels. The interaction between prostaglandins and the specific GPCRs depends on the differential expression of individual receptor subtypes in tissues and cells, their binding affinity for prostaglandins, and the differential activation of each receptor (19 -23). To date, there is no information on EP receptor subtypes or their specific functions in cholangiocarcinoma cells.
In light of recent evidence showing activation of EGFR by GPCRs (24 -26) and enhanced EGFR activation in cholangiocarcinoma cells (27,28), this study was designed to evaluate our hypothesis that the G-protein-coupled EP receptor may transactivate EGFR and that this mechanism may be important in cholangiocarcinogenesis. Our data reveal that COX-2derived PGE 2 transactivates EGFR through the EP 1 receptor in human cholangiocarcinoma cells and that this process involves the c-Src protein. Transactivation of EGFR subsequently induces Akt phosphorylation and enhances tumor cell proliferation and invasion. Furthermore, we show that activation of EGFR by EGF increases COX-2 expression, whereas blocking PGE 2 synthesis or the EP 1 receptor attenuates EGFinduced EGFR phosphorylation, suggesting that the COX-2/ PGE 2 /EP 1 receptor pathway also modulates activation of EGFR induced by its cognate ligand. Our findings reveal a novel interaction between COX-2-derived PGE 2 and EGFR signaling that synergistically promotes cancer cell growth and invasion.
Cell Culture and Transfections-Human cholangiocarcinoma cell lines, including CCLP1, SG231, and HuCCT1, were cultured according to our previously described methods (12,15,29). For transient transfection assays, the cultured cells were seeded at a concentration achieving 80% confluence in 6-well plates 18 h before transfection. The cells were transfected with the COX-2 expression plasmid (cloned into pcDNA) or the pcDNA control vector (1 g of plasmid for each transfection) using Lipofectamine Plus TM reagent. The cells with optimal overexpression of COX-2 were confirmed by immunoblotting and subsequently used for further experiments.
Cell Invasion Assay-The cell invasion assay was performed in Matrigel-coated Transwell chambers (BD Biosciences). Cells (4 ϫ 10 4 ) in 500 l of serum-free medium were seeded in the upper chamber in the presence or absence of different inhibitors or EP 1 receptor antagonist. Serum-free medium-containing vehicle, PGE 2 , or different EP receptor agonists were added to the lower chamber as chemoattractants. To determine the invasiveness of CCLP1 cells with antisense inhibition of EP receptors, the cells transfected with the antisense oligonucleotides for individual EP receptors or control cells were seeded in the upper chamber in serum-free medium, with the lower chamber containing vehicle or PGE 2 in serum-free medium. After 24 h of incubation at 37°C, the cells on the upper surface of the filter were mechanically removed with a cotton swab. The filter was fixed and stained using a Diff-Quik staining kit (Dade Behring Inc., Newark, DE) according to the manufacturer's instructions. The invading cells on the lower surface were counted under a microscope (magnification ϫ50). Five fields were counted per filter, and 4 wells were used for each treatment.
Phosphorylation of EGFR-CCLP1 cells were transfected with the COX-2 expression plasmid or EP 1 receptor small interfering RNA (siRNA) or treated with PGE 2 , EP 1 receptor agonist/antagonist, or EGFR inhibitors, and cell lysates were obtained. Equal amounts of the cell lysates were preincubated with 5 g/ml rabbit anti-human EGFR polyclonal antibody at 4°C, followed by the addition of 20 l of protein A/G-agarose (Santa Cruz Biotechnology, Inc.). The mixtures were incubated overnight at 4°C. After three washes with the same hypotonic buffer, the pellet was used for immunoblotting with anti-phosphotyrosine monoclonal antibody PY99.
Binding of EGFR to the EP 1 Receptor and c-Src-The binding complexes of EGFR and the EP 1 receptor and c-Src in CCLP1 cells were determined by immunoprecipitation and Western blotting. Confluent CCLP1 cells were serum-starved for 24 h and then treated with ONO-DI-004 or PGE 2 at 10 M for 30 min. Cell lysates were subsequently prepared for immunoprecipitation with antibody against the EP 1 receptor or c-Src, respectively. The immunoprecipitants were then subjected to SDS-PAGE and immunoblotting with anti-EGFR antibody.
RNA Interference (RNAi)-The sequence of EP 1 receptor siRNA was selected as described previously (18). The targeted sequence that effectively mediates the silencing of EP 1 receptor expression is 5Ј-AGCUUGUCG-GUAUCAUGGUTT-3Ј (sense). The 21-nucleotide synthetic EP 1 receptor siRNA duplex was prepared by Dharmacon, Inc. (Lafayette, CO). Cells were transfected with EP 1 receptor siRNA or a 21-nucleotide irrelevant RNA duplex as a control using Lipofectamine TM 2000. Depletion of the EP 1 receptor was confirmed by immunoblotting.
Gelatin Zymography-The matrix metalloprotease (MMP) proteolytic activity in the supernatants of the treated cells was analyzed for the level of MMP-2 by zymography. Briefly, an equal amount of serumfree medium from cells with different treatments was loaded onto 10% SDS-polyacrylamide gel containing 1 mg/ml gelatin (Invitrogen). After electrophoresis, SDS was removed from the gel by incubation in 2.5% Triton X-100 at room temperature for 30 min with gentle shaking. The gel was washed well with distilled water and incubated at 37°C for 16 -36 h in a developing buffer containing 50 mM Tris-HCl (pH 7.6), 0.2 M NaCl, 5 mM CaCl 2 , and 0.02% Brij 35. The gel was then stained with a solution of 30% methanol, 10% glacial acetic acid, and 0.5% Coomassie Blue G-250 and then destained in the same solution without dye. Proteinase activity was detected as unstained bands on a blue background representing areas of gelatin digestion.
Immunohistochemical Analysis for COX-2 and EGFR-Eleven archival formalin-fixed, paraffin-embedded specimens of human cholangiocarcinoma and surrounding non-tumor liver tissue were obtained from the University of Pittsburgh Medical Center. The tissue specimens were utilized for immunohistochemical analysis for COX-2 and EGFR following the protocol recommended by the University of Pittsburgh. None of the cases used in this study had patient identifiers, and strict confidentiality was maintained. For immunohistochemical staining of COX-2 and EGFR using human cholangiocarcinoma tissue, 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. The slides were incubated overnight at 4°C with 1:100 diluted anti-human COX-2 monoclonal antibody (obtained from Cayman Chemical Co., Inc.) and anti-EGFR antibody (obtained from Dako Corp.). Following repeated washings, the slides were incubated with biotin-conjugated secondary antibody (1: 200) for 30 min at room temperature. After probing, the avidin-peroxidase complex was added, and finally, 3,3Ј-diaminobenzidine substrate was utilized for color development. The slides were then counterstained with hematoxylin. The intensity of staining for COX-2 and EGFR was scored in each specimen on a scale of 0 -3, in which 0 ϭ negative staining, 1 ϭ weakly positive staining, 2 ϭ moderately positive staining, and 3 ϭ strongly positive staining. For each sample, 10 random high power fields were scored. The immunoreactivity for COX-2 and EGFR in each sequential section was documented and compared.
Inoculation of CCLP1 Cells into SCID Mice-Cultured CCLP1 cells were transfected with the COX-2 expression plasmid or pcDNA3 control vector, and the stably transfected cells were selected using G418. Cells (5 ϫ 10 6 ) suspended in phosphate-buffered saline were directly injected into the livers of SCID mice under anesthesia as described previously (33). After tumor cell implantation, the mice were kept under pathogenfree conditions, fed standard diet, and given free access to sterilized water. The mice were closely monitored for daily activity and killed 4 weeks after injection to document tumor growth. The tumor volume was calculated using the formula V ϭ Lϫ Wϫ Dϫ /6, where L is length, W is width, and D is depth of the tumor in millimeters. The animal protocol used followed the recommendations of the University of Pittsburgh Institutional Animal Care and Use Committee.

Expression of COX-2 and EGFR Is Increased in Human
Cholangiocarcinoma Tissues-Immunohistochemical stains were utilized to determine expression of COX-2 and EGFR in sequential sections of human cholangiocarcinoma and nonneoplastic bile ducts. Eleven paired cholangiocarcinomas and their matched non-tumor liver tissues were analyzed. The average COX-2 staining intensity in cholangiocarcinoma (2.10 Ϯ 0.54) was significantly higher than that in the non-neoplastic bile duct epithelium (0.64 Ϯ 0.10; p Ͻ 0.01). The average EGFR staining intensity in cholangiocarcinoma (1.45 Ϯ 0.92) was also significantly higher than that in the non-neoplastic bile duct epithelium (0.14 Ϯ 0.10; p Ͻ 0.01). Sequential sections from individual tissue specimens revealed cytoplasmic staining of COX-2 and membrane staining of EGFR in the same tumor cells (Fig. 1).
COX-2-derived PGE 2 Induces EGFR Phosphorylation in Cholangiocarcinoma Cells-Increased expression of COX-2 and EGFR in human cholangiocarcinoma cells suggests a possible interconnection between these two signaling pathways during cholangiocarcinogenesis. Given that COX-2-derived prostaglandins mediate effects primarily through specific plasma membrane receptors that are coupled with G-proteins and that certain GPCRs are known to activate EGFR (24 -26), we postulated that COX-2 may promote tumor growth through activation of EGFR. To evaluate this hypothesis, we first ex-amined the potential effect of COX-2 on phosphorylation of EGFR in cultured human cholangiocarcinoma cells. As shown in Fig. 2A, overexpression of COX-2 in CCLP1 cells enhanced EGFR phosphorylation. The COX-2-overexpressing cells exhibited increased PGE 2 production compared with the control vector cells (430 versus 295 pg/ml). Consistent with this, treatment of CCLP1 cells with PGE 2 induced rapid phosphorylation of EGFR (Fig. 2B). These findings indicate that COX-2-derived PGE 2 enhances EGFR phosphorylation in human cholangiocarcinoma cells.
The EP 1 Receptor Plays a Key Role in PGE 2 -induced EGFR Phosphorylation-Selective EP receptor subtype analogs were next utilized to determine their effects on PGE 2 -induced EGFR phosphorylation. As shown in Fig. 2C, treatment with the EP 1 receptor agonist ONO-DI-004 caused a significant increase in EGFR phosphorylation, an effect similar to that induced by PGE 2 . In addition, PGE 2 -induced EGFR phosphorylation was inhibited by the selective EP 1 receptor antagonist ONO-8711 as well as by the EGFR tyrosine kinase inhibitors AG 1478 and PD 153035. These findings provide pharmacological evidence for the involvement of the EP 1 receptor in PGE 2 -induced EGFR phosphorylation. The observation that PP2 partially prevented PGE 2 -induced EGFR phosphorylation suggests the involvement of the Src protein in this process.
The role of the EP 1 receptor in PGE 2 -induced EGFR phosphorylation was further examined by siRNA suppression of the EP 1 receptor. In this approach, CCLP1 cells transfected with EP 1 receptor siRNA or control RNA were treated with either Note the cytoplasmic staining of COX-2 and the membrane staining of EGFR in sequential sections of the same tumor. No stain was seen when the primary antibody was substituted with non-immunized serum. NC, negative control.

FIG. 2. PGE 2 induces EGFR phosphorylation via the EP 1 receptor in CCLP1 cells.
A, effect of COX-2 overexpression on EGFR phosphorylation. CCLP1 cells were transfected with the COX-2 expression plasmid in serum-free medium for 24 h. EGFR phosphorylation was determined by immunoprecipitation with anti-EGFR antibody and immunoblotting with anti-phosphotyrosine antibody (first panel). Total EGFR in the immunoprecipitate was determined by reprobing the same blot with anti-EGFR antibody (second panel). An equal amount of the same cell lysate was used for Western blotting to detect COX-2 expression (third panel), which was reprobed with actin (fourth panel). B, effect of PGE 2 on EGFR phosphorylation. CCLP1 cells cultured in serum-free medium for 24 h were treated with 10 M PGE 2 for the indicated times, and cell lysates were obtained. EGFR phosphorylation was determined by immunoprecipitation with anti-EGFR antibody, followed by immunoblotting with anti-phosphotyrosine antibody (upper panel). Total EGFR in the immunoprecipitates was determined by reprobing the same blot with anti-EGFR antibody (middle panel). Quantitative analysis of EGFR phosphorylation was performed by determining the ratio between the EGFR protein and phosphorylation levels from three different experiments (lower panel). *, p Ͻ 0.01 compared with the control. C, the EP 1 receptor antagonist ONO-8711 blocks PGE 2 -induced EGFR phosphorylation. CCLP1 cells were serum-starved for 24 h before being treated with ONO-8711, AG 1478, PD 153035, or PP2 for 30 min. The cells were subsequently treated with 10 M PGE 2 for 30 min, and cell lysates were obtained. EGFR phosphorylation was determined by immunoprecipitation with anti-EGFR antibody, followed by immunoblotting with anti-phosphotyrosine antibody (upper panel). Total EGFR in the immunoprecipitates was determined by reprobing the same blot with anti-EGFR antibody (middle panel). Quantitative analysis of EGFR phosphorylation was performed by determining the ratio between the EGFR protein and phosphorylation levels from three different experiments (lower panel). Increased EGFR phosphorylation was observed after treatment with ONO-DI-004 or PGE 2 at 10 M (*, p Ͻ 0.01 compared with the control). PGE 2 -induced EGFR phosphorylation was significantly blocked by pretreatment with the EP 1 receptor antagonist ONO-8711, the EGFR kinase inhibitor AG 1478 or PD 153035, or the c-Src inhibitor PP2 (**, p Ͻ 0.05 compared with PGE 2 treatment). D, RNAi suppression of the EP 1 receptor inhibits EGFR phosphorylation induced by PGE 2 and the EP 1 receptor agonist ONO-DI-004. CCLP1 cells were transfected overnight with EP 1 receptor siRNA or control RNA in serum-free medium and then treated with PGE 2 or ONO-DI-004 at 10 M for 30 min. EGFR phosphorylation was determined by immunoprecipitation with anti-EGFR antibody and immunoblotting with anti-phosphotyrosine antibody (upper panel). Total EGFR in the immunoprecipitates was determined by reprobing the same blot with anti-EGFR antibody (middle panel). Quantitative analysis of EGFR phosphorylation was carried out by calculating the ratio between the EGFR protein and phosphorylation levels from three different experiments (lower panel). RNAi suppression of EP 1 receptor expression significantly inhibited EGFR phosphorylation induced by either ONO-DI-004 (*, p Ͻ 0.05) or PGE 2 (**, p Ͻ 0.01). E, representative Western blot showing the level of EP 1 protein in CCLP1 cells transfected with EP 1 receptor siRNA or control RNA and non-transfected cells. PGE 2 or ONO-DI-004 at 10 M to determine EGFR phosphorylation. RNAi suppression of the EP 1 receptor significantly inhibited phosphorylation of EGFR induced by PGE 2 or the EP 1 receptor agonist ONO-DI-004 (Fig. 2D). The efficacy of EP 1 receptor depletion in this system was verified by Western blot analysis, showing successful reduction of EP 1 protein in cells transfected with EP 1 receptor siRNA (Fig. 2E).
Detection of the EP 1 Receptor-EGFR Complex in Human Cholangiocarcinoma Cells-To further examine the role of the EP 1 receptor in PGE 2 -induced EGFR activation, immunoprecipitation and Western blot experiments were performed to determine whether the EP 1 receptor associates with EGFR in cells.
In these experiments, CCLP1 cells were treated with the EP 1 receptor agonist ONO-DI-004 or PGE 2 at 10 M for 30 min, and cell lysates were then prepared and subjected to immunoprecipitation with anti-EP 1 receptor antibody, followed by immunoblotting with anti-EGFR antibody. Although a low level of the EP 1 receptor-EGFR complex was detected in the control cells, treat-ment with PGE 2 or ONO-DI-004 significantly enhanced the association between the EP 1 receptor and EGFR (Fig. 3A). In addition, PGE 2 or ONO-DI-004 also induced binding of EGFR to c-Src (Fig. 3B), suggesting the involvement of c-Src in EP 1 receptor-induced EGFR activation. These findings further support the involvement of the EP 1 receptor in EGFR phosphorylation.
EP 1 Receptor-mediated EGFR Activation Phosphorylates Akt-Given that Akt is activated by EGFR in other cell types (34), we next examined whether PGE 2 would induce Akt phosphorylation in CCLP1 cells. Increased Akt phosphorylation was observed in cells treated with either PGE 2 or the EP 1 receptor agonist ONO-DI-004 (Fig. 4). PGE 2 -induced Akt phosphorylation was blocked by the EP 1 receptor antagonist ONO-8711, the EGFR tyrosine kinase inhibitors AG 1478 and PD 153035, and the Src inhibitor PP2. These findings indicate that PGE 2 -induced Akt phosphorylation is mediated, at least in part, through activation of EGFR by the EP 1 receptor and that this process may involve the Src protein family.  ings, PGE 2 treatment of cultured CCLP1 cells significantly increased tumor invasion, MMP-2 activity, and cell proliferation in vitro (Fig. 5, B-D). Similarly, the EP 1 receptor agonist ONO-DI-004 also enhanced tumor cell invasion and proliferation. In contrast, PGE 2 -induced cell invasion, MMP-2 activity, and proliferation were attenuated by the EP 1 receptor antagonist ONO-8711. RNAi suppression of the EP 1 receptor also significantly inhibited PGE 2 -and EP 1 receptor agonist-induced cell invasion and proliferation (Fig. 6). Moreover, PGE 2 -induced cell invasion, MMP-2 activity, and proliferation were also inhibited by the EGFR tyrosine kinase inhibitors AG 1478 and PD 153035, by the Src inhibitor PP2, and by inhibition of Akt with LY 294002 (Fig. 5, B-D).

Involvement of the EP 1 Receptor, EGFR, and Akt in COX-2and PGE 2 -induced Cholangiocarcinoma Cell Growth and Inva
Effect of Other EP Receptor Subtypes on PGE 2 -induced Cholangiocarcinoma Cell Invasion-Because all four EP receptor subtypes (EP 1 , EP 2 , EP 3 , and EP 4 ) were detected in human cholangiocarcinoma cell lines CCLP1, HuCCT1, and SG231 as determined by reverse transcription-PCR (Fig. 7, A-C) and Western blotting (Fig. 7D), the potential involvement of these receptors in tumor cell invasion was determined using selective EP receptor subtype agonists and by antisense inhibition of individual EP receptors. As shown in Fig. 8A, the EP 2 receptor agonist butaprost, the EP 3 receptor agonist ONO-AE-248, and the EP 4 receptor agonist PGE 1 alcohol exhibited no effect on tumor cell invasion, whereas both PGE 2 and the EP 1 receptor agonist ONO-DI-004 increased tumor cell invasiveness under the same experimental conditions (*, p Ͻ 0.01 compared with the control).
For experiments with antisense inhibition of EP receptor subtypes, CCLP1 cells were transfected with the oligonucleotides specific for the EP 1 , EP 2 , EP 3 , and EP 4 receptors, and the transfected cells were then analyzed for PGE 2 -induced cell invasion. As shown in Fig. 8B, PGE 2 -induced cell invasion was blocked by antisense inhibition of the EP 1 receptor, but not by antisense inhibition of the EP 2 , EP 3 , or EP 4 receptor. Taken together, the data from the above pharmacological approaches with specific agonists and antagonists as well as molecular approaches with antisense and siRNA all support the involvement of the EP 1 (but not EP 2 , EP 3 , or EP 4 ) receptor in PGE 2induced cholangiocarcinoma cell invasion.
Activation of EGFR by EGF Increases COX-2 Expression and PGE 2 Production-The data presented above indicate that COX-2-derived PGE 2 activates EGFR/Akt through the G-protein-coupled EP 1 receptor and that this process is involved in cholangiocarcinoma cell growth and invasion. In light of the elevated expression of EGFR and COX-2 in human cholangiocarcinoma tissues (as shown in Fig. 1) and the known effect of receptor tyrosine kinase on COX-2 expression (9), we postulated that EGFR may reciprocally influence COX-2 expression and regulate cholangiocarcinoma cell growth. To examine this hypothesis, cultured CCLP1 cells were treated with EGF for 24 h, and cell lysates were obtained for Western blotting to determine the level of COX-2. As shown in Fig. 9A, EGF treatment significantly increased COX-2 expression; this effect appeared at 6 h and peaked at 24 -48 h. Consistent with this, EGF treatment enhanced the production of PGE 2 (Fig. 9B). EGF-induced PGE 2 release was blocked by the COX-2 inhibitor NS-398 as well as by the EGFR tyrosine kinase inhibitors AG 1478 and PD 153035; a similar effect was also observed with AACOCF 3 (Fig. 9B), a cytosolic phospholipase A 2 ␣ inhibitor that reduces the availability of arachidonic acid substrate for COX-2. Moreover, EGF-induced phosphorylation of EGFR was partially inhibited by pretreatment with AACOCF 3 , NS-398, ONO-8711, AG 1478, or PD 153035 (Fig. 9C). These findings indicate that EGFR activation further induces COX-2 expression and PGE 2 production and that this signaling is involved in activation of EGFR by its cognate ligand. DISCUSSION Several noteworthy findings are presented in this study. First, overexpression of COX-2 in human cholangiocarcinoma cells promoted tumor growth and invasion both in vitro and in a tumor xenograft model in SCID mice. Second, the expression and localization of COX-2 and EGFR in human cholangiocarcinoma cells was documented using immunohistochemical stains on sequential sections of human cholangiocarcinoma tissues. Third, this study provides the first evidence for transactivation of EGFR by COX-2 and PGE 2 in human cholangiocarcinoma cells and the key role of the EP 1 receptor in this process. Fourth, our findings demonstrate for the first time the key role of EP 1 receptor-mediated EGFR transactivation in Akt activation. Finally and most important, this study reveals a novel cross-talk between the COX-2-derived PGE 2 pathway and EGFR signaling for cancer cell growth and invasion, as illustrated in Fig. 10.
As PGE 2 exerts its bioactivity through four different G-protein-coupled EP receptors (EP 1 , EP 2 , EP 3 , and EP 4 ), it was important to evaluate their expression in cholangiocarcinoma cells. Although the mRNAs for all four EP receptor subtypes were detected in all three human cholangiocarcinoma cell lines (CCLP1, HuCCT1 and SG231), the EP 1 receptor mRNA is the most abundant form. Similarly, the EP 1 protein is also highly expressed in all three cell lines. Our findings in this study provide the first evidence for the expression profiles of EP receptors in human cholangiocarcinoma cells and depict a pivotal role of the EP 1 receptor in EGFR transactivation in human cancer cell proliferation and invasion. The latter conclusion is based on the following observations: 1) enhanced EGFR phosphorylation by COX-2 overexpression or PGE 2 treatment; 2) induction of EGFR phosphorylation by the selective EP 1 receptor agonist ONO-DI-004; 3) inhibition of PGE 2 -induced EGFR phosphorylation by the selective EP 1 receptor antagonist ONO-8711; 4) attenuation of PGE 2 -induced EGFR phosphorylation by EP 1 receptor siRNA; 5) increased binding of the EP 1 receptor to EGFR in response to PGE 2 or ONO-DI-004; 6) induction of Akt phosphorylation and cell invasion by ONO-DI-004, but not by the agonists for EP 2 , EP 3 and EP 4 ; 7) inhibition of PGE 2 -induced Akt phosphorylation, cell proliferation and invasion, and MMP-2 activity by ONO-8711 and the EGFR tyro- sine kinase inhibitors AG 1478 and PD 153035; and 8) inhibition of PGE 2 -induced cell invasion and proliferation by siRNA or antisense inhibition of the EP 1 (but not EP 2 , EP 3 , or EP 4 ) receptor.
Cross-talk between different members of receptor families has become a well established concept in signal transduction. GPCRs as well as receptor tyrosine kinases constitute prominent families of cell-surface proteins regulating the responsiveness of cells to environmental signals (35,36). Different classes of G-proteins have been shown to be involved in the transactivation of tyrosine kinase receptors, including the G i , G q , and G 13 proteins, although, to date, there have been no data available implicating G s -coupled receptors in EGFR signal transactivation (24,26). Consistent with these observations, in this study, we demonstrated a predominant role of the G q -coupled EP 1 receptor (but not the G s -coupled EP 2 and EP 4 receptors) in EGFR transactivation and cell proliferation/invasion. Given that the G s -coupled EP 2 and EP 4 receptors mediate their effect via increasing intracellular cAMP levels and that activation of cAMP/protein kinase A signaling is known to enhance the proliferation and motility of cholangiocytes and cholangiocarcinoma cells, the lack of EP 2 and EP 4 receptor effect revealed in this study further underscores the importance of EP 1 receptormediated EGFR transactivation in cholangiocarcinoma progression. It is of further interest that the experiments with the EP 3 receptor agonist ONO-AE-248 and EP 3 receptor siRNA also failed to show involvement of the G i -coupled EP 3 receptor, despite that fact that G i is known to exert its effect through reduction of cAMP.
The ability of GPCRs to transactivate EGFR can occur through several mechanisms, including extracellular release of EGF and other EGF-like ligands or through intracellular molecules, including Src family tyrosine kinases, and the inhibitory effects of reactive oxide species on EGFR-specific phosphatases (24 -26). Our data suggest that PGE 2 -induced EGFR transactivation in cholangiocarcinoma cells occurs through activation of c-Src. This interpretation is supported by the observations that PGE 2 or the EP 1 receptor agonist ONO-DI-004 enhanced formation of the Src-EGFR binding complex in CCLP1 cells and that inhibition of Src by PP2 prevented PGE 2induced Akt phosphorylation, tumor cell invasion, and MMP-2 activity. These findings are consistent with recent studies showing the involvement of Src in PGE 2 -induced transactivation of EGFR in colon cancer cells (37,38).
The EGFR family consists of four receptor tyrosine kinases, EGFR (ErbB1), ErbB2, ErbB3, and ErbB4. EGFR controls a wide variety of biological responses, such as proliferation, migration, and modulation of apoptosis, and the effects are mediated through activation of downstream molecules, including the phosphoinositide 3-kinase/Akt pathway (34). Aberrant EGFR signaling due to overexpression, mutation, or autocrine signaling loops has been implicated in several other human cancers (for review, see Refs. 39 -43). In this study, we showed that EGFR expression was increased in human cholangiocarcinoma tissue as determined by immunohistochemical analysis. Moreover, our data suggest the involvement of Akt in transducing the effects of EP 1 receptor-induced EGFR activation in human cholangiocarcinoma cells. The latter assertion is based on the findings that Akt was activated within 30 min after PGE 2 or EP 1 receptor agonist treatment and that this effect was inhibited by two EGFR kinase inhibitors, AG 1478 and PD 153035. These observations are further corroborated by additional findings that the EGFR kinase inhibitors prevented PGE 2 -induced cholangiocarcinoma cell proliferation and invasion and MMP-2 activity.
Akt plays a key role in tumorigenesis and cancer progression by stimulating cell proliferation and invasion or by inhibiting apoptosis (44 -46). Akt is composed of an N-terminal PH domain and a C-terminal kinase catalytic domain and is activated by a dual regulatory mechanism, including translocation to the were seeded in the upper chamber, and medium containing vehicle, PGE 2 , or EP 1 receptor agonist ONO-PI-004, the EP 2 receptor agonist butaprost, the EP 3 receptor agonist ONO-AE-248, or the EP 4 receptor agonist PGE 1 alcohol was added to the lower chamber. After 24 h, the cells on the upper surface of the filter were removed, and the filter was fixed and stained. The cells on the lower surface were counted under a microscope (magnification ϫ50). Five fields were counted per filter, and 4 wells were used for each treatment. The experiment was repeated three times. Whereas PGE 2 or ONO-DI-004 enhanced tumor cell invasion (*, p Ͻ 0.01 compared with the control), agonists for the EP 2 , EP 3 , and EP 4 receptors exhibited no effect. B, effect of antisense inhibition of EP receptor subtypes on PGE 2 -induced cholangiocarcinoma cell invasion. CCLP1 cells (4 ϫ 10 4 ) transfected with antisense (As) oligonucleotides for the EP 1 , EP 2 , EP 3 , and EP 4 receptors were seeded in the upper chamber; medium containing PGE 2 or vehicle was added to the lower chamber. After 24 h, the cells on the upper surface of the filter were removed, and the filter was fixed and stained. The cells on the lower surface were counted under a microscope. Five fields were counted per filter, and 4 wells were used for each treatment. The data represent the average results from three experiments. PGE 2 -induced CCLP1 cell invasion was blocked by antisense inhibition of the EP 1 receptor, but not by antisense inhibition of the EP2, EP3, or EP4 receptor. *, p Ͻ 0.01 compared with the control; **, p Ͻ 0.05 compared with PGE 2 treatment of cells transfected with the control oligonucleotide. C, Western blots showing successful depletion of the EP 1 , EP 3 , and EP 4 receptors in CCLP1 cells transfected with the corresponding antisense oligonucleotides. Western blotting for the EP 2 receptor was not performed in EP 2 receptor antisense cells because of the low basal expression of the EP 2 receptor in CCLP1 cells.
plasma membrane and phosphorylation. The generation of phosphatidylinositol 3,4,5-triphosphate on the inner layer of the plasma membrane following phosphatidylinositol 3-kinase activation recruits Akt by direct interaction with its PH domain. At the membrane, Akt is phosphorylated by 3-phosphoinositide-dependent protein kinase-1 and -2. Phosphorylated Akt dissociates from the membrane and enters the cytoplasm and nucleus, where it phosphorylates several key proteins mediating cellular effects, such as stimulation of cell cycle progress and invasiveness. Increased expression of phosphorylated Akt has recently been documented in human cholangiocarcinoma cells by immunohistochemical staining of human cholangiocarcinoma tissues (14). The latter observation and the findings that the COX-2 inhibitor celecoxib inhibits Akt phosphorylation in cultured cholangiocarcinoma cells (14,16,17) suggest a possible involvement of Akt activation in COX-2-mediated cholangiocarcinoma cell growth, although a direct effect of COX-2-derived prostaglandin on Akt phosphorylation was not demonstrated prior to this study in light of the presence of COX-2-independent effects associated with celecoxib. Our present data establish a direct effect of PGE 2 on Akt phosphorylation in human cholangiocarcinoma cells. To our knowledge, this is the first study detailing the role of EP receptor subtype in Akt phosphorylation and EGFR transactivation in human cancer cells.
In addition to EGFR transactivation by COX-2-derived PGE 2 , our data also show that treatment of human cholangiocarcinoma cells with EGF increased COX-2 expression and PGE 2 synthesis, suggesting the involvement of EGFR in COX-2 FIG. 9. EGF treatment increases COX-2 expression and PGE 2 production in CCLP1 cells. Blocking PGE 2 signaling partially inhibited the phosphorylation of EGFR induced by EGF. A, EGF induces COX-2 expression. Cultured CCLP1 cells were serum-starved for 24 h and then treated with 20 ng/ml EGF for the indicated times. Cell lysates were obtained for Western blotting to determine the level of COX-2. EGF treatment increased COX-2 expression. This effect was observed at 6 h and peaked at 24 -48 h. B, EGF increases PGE 2 production. The effects of the cytosolic phospholipase A 2 ␣ inhibitor AACOCF 3 , the COX-2 inhibitor NS-398, the c-Src family inhibitor PP2, and the EGFR tyrosine kinase inhibitors AG 1478 and PD 153035 were determined. Serum-starved CCLP1 cells were pretreated with AACOCF 3 or NS-398 for 2 h or with PP2, AG 1478, or PD 153035 for 30 min. The cells were then continuously treated with 20 ng/ml EGFR for 16 h, and the media were subsequently collected to measure PGE 2 production using the PGE 2 enzyme immunoassay system (Amersham Biosciences) according to the manufacturer's protocol. EGF treatment significantly increased PGE 2 production (*, p Ͻ 0.01 compared with the control); this effect was attenuated by AACOCF 3 , NS-398, AG 1478, PD 153035, or PP2 (**, p Ͻ 0.05 compared with EGF treatment alone). The results are presented as the mean Ϯ S.D. of four experiments. C, the cytosolic phospholipase A 2 ␣ inhibitor AACOCF 3 , the COX-2 inhibitor NS-398, the EP 1 receptor antagonist ONO-8711, and the EGFR tyrosine kinase inhibitors AG 1478 and PD 153035 partially inhibit EGF-induced phosphorylation of EGFR. CCLP1 cells were serum-starved for 24 h and then pretreated with AACOCF 3 or NS-398 for 2 h or with ONO-8711, AG 1478, or PD 153035 for 30 min. The cells were subsequently treated with 20 ng/ml EGFR for 30 min, and cell lysates were obtained to determine EGFR phosphorylation. EGFR phosphorylation induced by EGF was partially   FIG. 10. Proposed mechanisms for COX-2-and PGE 2 -mediated cholangiocarcinoma cell invasion and proliferation. Prostaglandin synthesis is controlled by coupled activation of cytosolic phospholipase A 2 ␣ (cPLA 2 ␣) and COX-2 along the cell membrane. The produced prostaglandins are released to the extracellular space by the prostaglandin transporter. The major prostaglandin in biliary epithelial and cancer cells is PGE 2 . After PGE 2 is released into the extracellular space, it binds to the membrane G-protein-coupled EP receptors on the same cell (autocrine) or on a neighboring cell (paracrine). Although there are four types of PGE 2 receptors (EP 1 , EP 2 , EP 3 , and EP 4 ), our data show that the EP 1 receptor plays a key role in COX-2-and PGE 2 -mediated cholangiocarcinoma cell invasion and proliferation and that this effect is mediated, at least in part, by activation of EGFR/Akt. On the other hand, activation of EGFR further enhances COX-2 expression and PGE 2 production, which further amplify COX-2/EP 1 /EGFR/ Akt signaling. The PGE 2 /EP 1 receptor-induced transactivation of EGFR in cholangiocarcinoma cells is mediated, at least in part, through the c-Src-mediated intracellular mechanism; it remains unknown whether this process also involves extracellular release of endogenous EGFR ligand. AA, arachidonic acid.
blocked by pretreatment with AACOCF 3 , NS-398, ONO-8711, AG 1478, or PD 153035. *, p Ͻ 0.01 compared with the control; **, p Ͻ 0.05 compared with EGF treatment; ***, p Ͻ 0.01 compared with EGF treatment. expression. In this context, our result is consistent with that of Yoon et al. (11), who showed the involvement of EGFR in bile salt-induced COX-2 expression in cholangiocytes and cholangiocarcinoma cells. Collectively, these findings suggest that COX-2-derived PGE 2 signaling may mediate the actions of EGFR in cholangiocarcinogenesis. This assertion is further supported by the observations that EGFR phosphorylation induced by its cognate ligand (EGF) was partially inhibited by blocking PGE 2 synthesis with the cytosolic phospholipase A 2 ␣ inhibitor AACOCF 3 and the COX-2 inhibitor NS-398 or with the selective EP 1 receptor antagonist ONO-8711.
In summary, this study has established a novel EP 1 receptormediated transactivation of EGFR by COX-2-derived PGE 2 , which is crucial for COX-2-and PGE 2 -induced cholangiocarcinoma cell growth and invasion. Moreover, we have shown that activation of EGFR further up-regulates COX-2 expression and thus enhances EGFR signaling via activation of the EP 1 receptor. The cross-talk between these two key signaling systems likely plays an important role in COX-2-and receptor tyrosine kinase-induced cholangiocarcinogenesis. Given the recently reported side effect associated with the currently available COX-2 inhibitors in patients (47,48), our findings suggest that combinational utilization of agents targeting the EP 1 receptor and EGFR may represent a promising cancer therapeutic strategy that deserves further investigation.