Extracellular Matrix Fibronectin Increases Prostaglandin E2 Receptor Subtype EP4 in Lung Carcinoma Cells through Multiple Signaling Pathways

We have previously demonstrated that fibronectin (Fn) stimulates the proliferation of non-small cell lung carcinoma (NSCLC) cell growth through the induction of cyclooxygenase-2 (COX-2) and prostaglandin E2 secretion. Here, we demonstrate that NSCLC cells express mRNA and protein for the prostaglandin E2 receptor EP4 and that Fn enhances its stimulatory effect by inducing the expression of EP4, but not of EP1, EP2, and EP3 receptor subtypes. The effect of Fn on EP4 was inhibited by an antibody against α5β1 integrin and by inhibitors of phosphoinositide 3-kinase (wortmannin) and extracellular signal-regulated kinase (PD98095), but not by inhibitors of protein kinase C (calphostin C), of protein kinase A (H-89), or of mammalian target of rapamycin (rapamycin). A COX-2 small interfering RNA was also inhibitory. Fn significantly increased AP-2 binding activity in the promoter of the EP4 gene, and AP-2 antisense oligonucleotides blocked Fn-induced EP4 expression. Using full-length and mutated EP4 promoter constructs, we found that Fn stimulation of EP4 gene expression was inhibited when one AP-2 site (–1000 bp) was mutated. Fn induced nuclear AP-2α protein expression through multiple signaling pathways. Our results indicate that Fn-induced NSCLC cell proliferation is mediated through EP4. Furthermore, they show that Fn induces EP4 expression through the activation of α5β1-dependent signals that include induction of extracellular signal-regulated kinase and phosphoinositide 3-kinase pathways as well as expression of COX-2. These events lead to activation of the transcription factor AP-2α, which interacts with specific regions in the EP4 gene promoter, leading to transcription of the EP4 gene.

Extracellular matrix proteins are considered to play roles in the migration and differentiation of various cells, including car-cinoma cells. Fibronectin (Fn), 2 a matrix glycoprotein highly expressed in tobacco-related lung diseases, has been shown to stimulate carcinoma cell growth, including lung carcinoma (1)(2)(3). We previously reported that Fn stimulated human lung carcinoma cell growth through induction of cyclooxygenase-2 (COX-2) signaling and subsequent production of prostaglandin E 2 (PGE 2 ) (4). High PGE 2 levels are considered important in carcinoma growth and progression, and inhibition of PGE 2 synthesis blocks growth of carcinoma cells (5). The effect of PGE 2 has been attributed to its known capacity to bind to its receptors, designated EP1, EP2, EP3, and EP4 (6). These receptors have been implicated in carcinoma cell growth and progression (5,7,8,9). PGE 2 , acting via EP4, contributes to tumor growth and progression of gallbladder and colorectal carcinoma (7,8). A recent study of EP4 knock-out mice suggested a role for EP4 receptor in colon carcinoma (7). PGE 2 and its signaling through the EP4 receptor have been shown to mediate non-small cell lung carcinoma (NSCLC) invasiveness (10). Taken together, these observations suggest that manipulation of prostaglandin metabolism downstream from COX-2 produces more profound effects on carcinoma reduction than COX-2 inhibition alone and could be the basis for new approaches for the prevention of carcinoma.
At present, the mechanisms that link Fn and EP4 gene expression and how they might relate to lung carcinoma are unknown. Herein, we explore the relationship between these molecules and their role in lung carcinoma cell growth. Our results show that Fn stimulates lung carcinoma cell growth and that this inductive effect is partly dependent upon stimulation of PGE 2 production and induced PGE 2 receptor subtype EP4 gene expression, which is mediated through integrin-dependent signals, including the activation of phosphoinositide 3-kinase (PI3K) and extracellular signal-regulated kinase (ERK) signaling pathways. * This work was supported by American Lung Association Grant RG-10215N (to S. W. H.) and by a Merit Review Grant from the Department of Veterans Affairs (to J. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom all correspondence and reprint requests should be addressed: Division extension incubation of 7 min at 72°C. Analysis of amplicons was accomplished on 1% agarose gel containing 0.2 g/l ethidium bromide and visualized under UV transilluminator. The densitometric analysis of PCR products was performed by computer software (Bio-Rad Quantity One) and a GS-800 Imaging Densitometer (Bio-Rad) and standardized to the GAPDH product. EP4/GAPDH density bands in control groups were considered as 100%. Values of treatment group EP4/GAPDH ratios are given as percentage of controls. A 100base pair ladder (Invitrogen) was used as a size standard.
Real Time RT-PCR-This procedure, which is based on the time point during cycling when amplification of the PCR product is first detected, rather than on the amount of PCR product accumulated after a fixed number of cycles, was described previously (12). Final results, which were expressed as n-fold differences in EP4 gene expression relative to the GAPDH gene, were calculated using a formula based on a doubling of the product after each cycle (12). The procedures for treatment and total RNA preparation were identical to those described for RT-PCR. All PCRs using the LightCycler-FastStart DNA Master SYBR Green I kit were performed in the Cepheid Smart-Cycler real time PCR cycler (Sunnyvale, CA). The cycling conditions were as follows: initial denaturation at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s, 60°C for 10 s, and 72°C for 10 s. Experiments were performed in triplicate for each data point. For all experiments, controls without templates were included.
Oligodeoxynucleotides and Small Interfering RNA (siRNA) Treatments-The sequences of phosphorothioate oligodeoxynucleotides (ODN) for ERK and AP-2 used in this study were as follows. The antisense ODN sequence was 5Ј-GCCGCCGCC-GCCGCCAT-3Ј (referred to as AS ERK2), and the corresponding sense ODN sequence was 5Ј-ATGGCGGCGGCGGCGGC-3Ј (referred to as S ERK2). AP-2 antisense was 5Ј-CGTCAATTTC-CAAAGCATTTTCATGGATCGG-3Ј (Ϫ13 to ϩ18 of the AP-2 sequence), and the control oligonucleotide, 5Ј-CAAAGTCT-TGCATTATTCGGTCATAATGGCC-3Ј, consisted of similar nucleotides with scrambled sequences. They were synthesized by Sigma (The Woodlands, TX) according to published data (13,14). The COX-2 siRNA (catalog ID number 116912) was purchased from Ambion. EP4 siRNA (catalog number M-005714-00) and control nonspecific siRNA oligonucleotides (catalog number D-001206-13-05) were purchased from Dharmacon, Inc. (Lafayette, CO), as described previously (15). For the transfection of ERK and AP-2 ODN, cells were grown to 70% confluence, and a 1 M concentration of ERK and 4 M AP-2 phosphorothioate ODN mixed with FuGENE 6 transfection reagent per well of serum-free medium was added to the cells for 24 h at 37°C. COX-2 and EP4 siRNA or control siRNA were transfected using the oligofectamine reagent (Invitrogen) according to the manufacturer's instructions. Briefly, oligofectamine reagent was incubated with serum-free medium for 10 min. Subsequently, a mixture of respective antisense or sense ODN or siRNA was added. After incubation for 15 min at room temperature, the mixture was diluted with medium and added to each well. The final concentration of siR-NAs in each well was 100 nM. After culturing for 30 h, cells were washed and resuspended in new culture medium in the presence 16,2007 or absence of Fn for an additional 24 h for Western blot analysis, cell growth, and gel mobility shift assays.
Cell Viability Assay-NSCLC cells (10 5 cells/well) were transfected with EP4 siRNA for 30 h before exposing the cells to Fn (20 g/ml)-coated culture plates or dmPGE 2 (0.1 M) for an additional 48 h in 96-well plates. Afterward, the numbers of viable cells in culture were determined using the CellTiter-Glo luminescent cell viability assay kit, which is based on quantitation of ATP, an indicator of metabolically active cells, according to the manufacturer's instructions (Promega).
[methyl-3 H]Thymidine Incorporation Assay-This method was described previously (17). Briefly, human NSCLC cells were transfected with EP4 or control siRNAs for 30 h or with AP-2 antisense or sense ODN for 24 h before incubating with 1 Ci/ml [methyl-3 H]thymidine (specific activity 250 Ci/mmol; Amersham Biosciences) and exposing the cells to Fn (20 g/ml)-coated culture plates or dmPGE 2 (0.1 M) for an additional 24 h. Medium was removed, and attached cells were washed with 1ϫ PBS. Afterward, the attached cells were treated with ice-cold 6% trichloroacetic acid at 4°C for 20 min and washed once with 6% trichloroacetic acid. Cells were solubilized with 0.1 N NaOH and counted in a liquid scintillation counter in 4 ml of scintillation fluid.
Transient Transfection Assay-The human EP4 wild-type and deletion promoter constructs (pGlep4-1 to -5) ligated to the luciferase reporter gene have been reported previously (18). The EP 4 promoter construct contains ϳ4200 bp of the 5Ј-flanking region of the mouse EP4 receptor gene connected to the pGL3 basic luciferase reporter vector (Promega). NSCLC cells were seeded at a density of 5 ϫ 10 5 cells/well in 6-well dishes and grown to 50 -60% confluence. For each well, 2 g of the above plasmid DNA constructs and 0.2 g of the internal control phRL-TK Synthetic Renilla luciferase reporter vector were cotransfected into the cells using FUGENE 6 lipofection reagent (Roche Applied Science) as described in our earlier work (19). After 24 h of incubation, cells were treated with or without PD98095 (25 M) or wortmannin (100 nM) for 1 h before exposing the cells to Fn-or collagen type 1 (20 g/ml each)-coated culture plates. The preparation of cell extracts and measurement of luciferase activities were carried out using the dual luciferase reporter kit according to recommendations by the manufacturer (Promega). The assays for firefly luciferase activity and Renilla luciferase activity were performed sequentially in a Labsystems Luminoskan Ascent luminometer equipped with dual injectors. Changes in firefly luciferase activity were calculated and plotted after normalization with changes in Renilla luciferase activity within the same sample.
Statistical Analysis-All experiments were repeated a minimum of three times. All data from cell growth assays, gel shift assays, luciferase activity assays, RT-PCR or real-time RT-PCR, and Western blot analysis were expressed as mean Ϯ S.D. The data presented in some figures are from a representative experiment, which was qualitatively similar in the replicate experiments. Statistical significance was determined with Student's t test (two-tailed) comparison between two groups of data sets. The asterisks shown in the figures indicate significant differences of experimental groups in comparison with the corresponding control condition (p Ͻ 0.05; see figure legends).

Effect of Fn on EP4 Gene Expression in Human Lung
Carcinoma Cells-We previously showed that mRNAs encoding for the four PGE 2 receptor subtypes are present in human NSCLC cells (17). Consistent with this, we show that the PGE 2 receptor subtype EP4 protein is expressed in the two NSCLC cell lines studied (Fig. 1A). We also demonstrated that Fn stimulated NSCLC cell growth (4). Here, we examined whether blockade of EP4 could influence the effects of Fn on cell growth. We first depleted EP4 from cells in culture using siRNA approaches. Treatment of H1838 cells with EP4 siRNA blocked EP4 production. Levels of EP4 were unchanged in cells transfected with control siRNA oligonucleotides (Fig. 1B). To determine if EP4 siRNA can block Fn-induced lung carcinoma cell growth in our system, H1838 cells were transfected with EP4 siRNA duplexes. Afterward, the cells were plated onto Fn-coated culture plates for an additional 24 h. As shown in Fig. 1C, the EP4 siRNA duplexes inhibited Fn-induced H1838 cell proliferation, whereas the control siRNA had no effects as determined by the [ 3 H]thymidine incorporation assay. Similar results were also found by cell viability assays (Fig. 1D).
Since EP4 has been shown to be involved in human lung carcinoma biology, we tested if Fn can affect its expression. H1838 cells exposed to Fn showed increased EP4 protein levels in a time-and dose-dependent manner with maximal increases noted in 24 h at concentrations of 20 g/ml (Fig. 2, A and C). Similar results were also found in an additional NSCLC cell line (H2106) (Fig. 1, B and D). Fn also significantly stimulated EP4 mRNA levels in a dose-dependent manner, with maximal increases noted at a concentration of 20 g/ml Fn as determined by RT-PCR (Fig. 2E), whereas collagen type 1, another matrix glycoprotein, had no effect (Fig. 2F). This result was confirmed by real time RT-PCR analysis (Fig. 2G). Of note, cells cultured on Fn-and collagen type 1-coated plates adhered well and showed similar morphology (not shown). In order to determine the specificity of the effect of Fn on EP4, we examined whether Fn can affect other PGE 2 receptor subtypes. We found that increased concentrations of Fn had no effect on other EP receptors (EP1, EP2, and EP3) (Fig. 2H); nor did silencing EP4 gene expression alter the effect of Fn on these EP receptors (Fig.  2I). Similar results were obtained with H2106 cells (not shown).
␣5␤1 Integrin, PI3K, ERK, and COX-2, but Not PKC and PKA Signaling Are Involved in Fn-induced EP4 Expression-We then determined if the effect of Fn was mediated by its integrin receptor ␣5␤1. To this end, we treated H1838 cells with or without anti-integrin ␣5␤1 (MAB1969) or anti-integrin ␣2␤1 antibodies (MAB1967; 25 g/ml each) for 2 h before exposing the cells to culture plates coated with Fn. We found that Fninduced EP4 protein was eliminated in the presence of anti-␣5␤1 antibodies, whereas the anti-␣2␤1 antibodies had no effect (Fig. 3A). This suggests that ␣5␤1 integrin mediates Fninduced EP4 expression.
Fn has been shown to affect kinase signaling pathways in several studies (21)(22)(23). We previously demonstrated that Fn activated ERK and PI3K/Akt signaling pathways in NSCLC cells (4,23,24). Here, we examined if inhibition of these kinase signal pathways diminished or abrogated the effects of Fn on EP4. The specific inhibitors of ERK (PD98095 (25 M)) and of PI3K (wortmannin (100 nM)) significantly blocked Fn-induced EP4 protein levels in H1838 cells (Fig. 3B). We previously demonstrated that mTOR signals were mediating some of the effects of Fn on NSCLC cell growth (23). However, we found that rapamycin, an inhibitor of mTOR, had no effect on inhibition of Fn-induced expression of EP4 protein (Fig. 3C). Also, the inhibitor of PKA (H89 (10 M)) or of PKC (calphostin C (Cal; 0.5 M)) had no effects (Fig. 3, D and E). In addition, we showed

Fibronectin Stimulates EP4 in Lung Cancer
that the inhibitor of PI3K, wortmannin, blocked the effect of Fn on the phosphorylation of ERK (Fig. 3F), whereas the inhibitor of ERK, PD98095, had no effect on Fn-induced phosphorylation of Akt (Fig. 3G). This suggests that ERK is downstream of the PI3K/Akt pathway. Similar results were obtained with H2106 cells (not shown).
Fn Increased EP4 Promoter Activity-We next examined whether the effects of Fn on EP4 expression occur at the transcriptional level. The EP4 promoter contains multiple transcription factor binding sites, including NF-B, NF-IL6 (C/EBP), Sp1, and AP-2, among others (Fig. 4A). These sites have been shown to be differentially responsive to various stimuli (15,18,25,26). We found that H1838 cells, transfected with the full-length wild-type EP4 promoter (Ϫ4200/Ϫ116 bp) luciferase reporter construct, exposed to Fn showed increased promoter activity. Collagen type 1 had no effect on the wild-type promoter (Fig. 4B). The Fn-induced EP4 promoter activity was slightly reduced in one EP4 deletion reporter construct (Ϫ1555/Ϫ116 bp). There was no response to Fn with another EP4 deletion reporter construct (Ϫ992/Ϫ116 bp) (Fig. 4B), indicating that the region between Ϫ1555 and Ϫ992 bp in the EP4 promoter played an important role in stimulation of EP4 gene expression in response to Fn. We also tested whether kinase signal pathways were involved in EP4 transcriptional regulation by Fn and found that, as expected, the inhibitors of either ERK or PI3K partially reduced Fn-induced EP4 promoter activities (Fig. 4C). Similar results were obtained with H2106 cells (not shown).
Regulation of AP-2, Sp1, and C/EBP in the EP4 Promoter by Fn in Human Lung Carcinoma Cells-To further explore the role of Fn in regulation of EP4 promoter activity, electrophoretic mobility shift assays were performed to identify the transcription factors involved. We found that H1838 cells treated with Fn for 24 h showed a significant increase in AP-2 ( Fig. 5C), but we observed no effect in C/EBP (Fig. 5A) and a slight increase in Sp1 (Fig. 5B) nuclear protein binding activities as compared with solvent controls. In contrast, collagen type 1 (20 g/ml) had no effect. We also tested if the activation of ERK signals by Fn was involved in the induction of AP-2 binding activity. ERK1/2 antisense oligonucleotide (1 M) completely blocked ERK1 and reduced ERK2 production. The levels of ERK1, ERK2, and actin were unaffected in untransfected cells and in cells treated with ERK sense oligonucleotides (Fig. 5D,  top). The antisense ERK ODN prevented the Fn-induced AP-2 binding activity (Fig. 5D, bottom). Similarly, the inhibitors of ERK (PD98095) and of PI3K (wortmannin) also blocked Fn- E and F, dose-dependent effect of Fn and collagen on EP4 mRNA expression. Total RNA was isolated from H1838 cells cultured on plates coated with increased concentrations of Fn or collagen and then subjected to RT-PCR analysis. GAPDH expression was evaluated as internal control. G, effect of Fn on EP4 mRNA expression as determined by real time RT-PCR. Total RNA was isolated from H1838 cells cultured in the presence or absence of Fn or collagen type 1 (20 g/ml each), followed by real time PCR analysis. GAPDH served as internal control for normalization purposes. *, significant differences from control (p Ͻ 0.05). Con, indicates untreated control cells. H, dose-dependent effect of Fn on EP1, EP2, and EP3 expression. Cellular proteins were isolated from H1838 cells cultured on plates coated with increasing concentrations of Fn as indicated, followed by Western blot for EP1, EP2, and EP3 protein levels. I, time-dependent effect of Fn on EP1, EP2, and EP3 expression. Cellular proteins were isolated from H1838 cells transfected with control or EP4 siRNA pool for 30 h before exposing the cells to culture plates coated with Fn (20 g/ml) for the indicated time period. Afterward, Western blot was performed to detect EP1, EP2, and EP3 protein levels. Actin served as internal control for normalization purposes. MARCH 16, 2007 • VOLUME 282 • NUMBER 11

JOURNAL OF BIOLOGICAL CHEMISTRY 7965
induced AP-2 binding activity (Fig. 5E). The addition of an AP-2␣ antibody induced a supershift band, whereas AP-2␤ and AP-2␥ antibodies had no effects (Fig. 5F). The specific bands for AP-2, Sp1, and C/EBP were attenuated by a 100-fold molar excess of unlabeled wild-type oligonucleotides but were not inhibited by the mutated unlabeled oligonucleotides (Fig. 5, Mut). Oligonucleotides containing a mutated AP-2 (Mut AP-2), Sp1 (Mut Sp1), or C/EBP (Mut C/EBP) site were end-labeled with [␥-32 P]ATP and used as another control to confirm the binding specificity. Similar results were obtained with H2106 cells (not shown).

The Role of Transcription Factor AP-2 in Fn Induction of EP4 and
Cell Growth-We further tested the role of AP-2 in mediating Fn-induced EP4 expression in human lung carcinoma cells by using the antisense approach. We showed that AP-2 antisense ODN completely abrogated the production of AP-2. Furthermore, this antisense ODN eliminated the Fn-induced EP4 protein production, whereas the AP-2 sense ODN had no effect (Fig. 6A). Consistent with these findings, we found that cells transfected with AP-2 antisense ODN resulted in inhibition of Fn-induced cell growth as determined by [methyl-3 H]thymidine incorporation assay (Fig. 6B) and inhibition of Fn-stimulated effect on EP4 promoter activities (Fig. 6C). The control sense ODN had no effect. By using site-directed mutated EP4 promoter constructs in which each of three AP-2 binding sites were mutated (Ϫ1529, Ϫ1133, and Ϫ1000 bp), we found that the stimulatory effect of Fn on EP4 promoter activity was lost with EP4 promoter constructs in which one AP-2 site was mutated (Ϫ1000 bp) (Fig. 6D).
PI3K, ERK, and ␣5␤1 Integrin Signaling Are Involved in Fn-induced AP-2 Protein Expression-We also examined the effects of Fn on AP-2 protein expression, and, consistent with the gel shift experiment results, FIGURE 3. Involvement of ␣5␤1 integrin, PI3K, and ERK pathways in the induction of EP4 by Fn. A, effect of anti-␣5␤1 antibodies on Fn-induced EP4 protein levels. Cellular protein was isolated from H1838 cells cultured for up to 2 h in the presence or absence of anti-␣5␤1 and anti-␣2␤1 antibodies (25 g/ml each) before exposing the cells to culture plates coated with Fn for an additional 24 h and then subjected to Western blot analysis for EP4. B, effect of inhibitors of ERK1/2 and PI3K on Fn-induced EP4 protein levels. Cellular protein was isolated from H1838 cells cultured for up to 2 h in the presence or absence of PD98095 (25 M) or wortmannin (100 nM) before exposing the cells to Fn-coated culture plates for an additional 24 h and then subjected to Western blot analysis for EP4. C, effect of mTOR inhibitor on Fn-induced EP4 protein levels. Cellular protein was isolated from H1838 cells cultured for up to 4 h in the presence or absence of rapamycin (10 nM) before exposing the cells to Fn-coated culture plates for an additional 24 h and then subjected to Western blot analysis for EP4 protein. D, effect of PKA inhibitors on Fn-induced EP4 protein levels. Cellular protein was isolated from H1838 cells cultured for up to 2 h in the presence or absence of H89 (10 M) before exposing the cells to Fn-coated culture plates for an additional 24 h and then subjected to Western blot analysis. E, effect of PKC inhibitors on Fn-induced EP4 protein levels. Cellular protein was isolated from H1838 cells cultured for up to 2 h in the presence or absence of calphostin C (0.5 M) before exposing the cells to Fn-coated culture plates for an additional 24 h, then subjected to Western blot analysis. F, effect of PI3K inhibitor on ERK1/2. Cellular protein was isolated from H1838 cells cultured for 1h in the presence or absence of wortmannin (100 nM) before exposing the cells to Fn (20 g/ml) coated onto the culture plates for an additional 1 h. Afterward, Western blot was performed to detect phosphorylated ERK1/2 and total ERK1 and ERK2 proteins. G, effect of ERK inhibitor on Akt. Cellular protein was isolated from H1838 cells cultured for 1 h in the presence or absence of PD98095 (25 M) before exposing the cells to Fn (20 g/ml) for an additional 1 h. Afterward, Western blot was performed to detect phosphorylated Akt and total Akt protein. Actin served as internal control for normalization purposes. Con, untreated control cells.

COX-2 Signaling Is Involved in Fn-induced AP-2 and EP4
Protein Expression-Having established that Fn stimulates NSCLC cell proliferation through COX-2 (4) and via EP4 (this study), we explored whether COX-2 signals could also affect EP4 expression. We found that COX-2 siRNA blocks endogenous COX-2 protein production; no changes were noted in cells transfected with control siRNA (Fig. 8A). COX-2 siRNA significantly abrogated Fn-induced EP4 expression; the control siRNA had no effect (Fig. 8A). We also found that, under the same conditions, COX-2 siRNA abrogated the effect of Fn on AP-2␣ protein expression (Fig. 8B), whereas the control siRNA had no effect. These studies suggest that COX-2 activation not only promotes PGE 2 secretion but may also amplify its effects by stimulating EP4 expression. Therefore, we tested whether PGE 2 can stimulate EP4 expression. We observed that dmPGE 2 (0.1 M) also enhanced AP-2␣ and EP4 protein expression, with greater stimulation seen when Fn and dmPGE 2 were used concomitantly (Fig. 8, C and D). Finally, similar to the effect of Fn on cell growth shown in Fig. 1, C and D, the stimulatory effect of PGE 2 on human cancer cell proliferation was diminished in the presence of EP4 siRNA, whereas the control siRNA had no effect as determined by the [ 3 H]thymidine incorporation assay (Fig. 8E) and by cell viability assays (Fig. 8F). Similar results were obtained with H2106 cells (not shown).

DISCUSSION
Fn is a heterodimeric extracellular matrix glycoprotein implicated in a number of physiological events during embryogenesis, angiogenesis, thrombosis, and inflammation (27)(28)(29). Fn expression is increased in lung carcinomas, particularly in non-small cell lung carcinoma (3, 29 -31). Also, the adhesion of lung carcinoma cells to Fn enhances tumorigenicity and confers resistance to apoptosis induced by standard chemotherapeutic agents (32). Previously, we found that Fn stimulates human lung carcinoma cell growth in vitro by increasing expression of COX-2 and PGE 2 biosynthesis (4). We also demonstrated that all four PGE 2 receptors are expressed in NSCLC cells studied (17). Based on these data, we predicted that Fn stimulated NSCLC growth through one or more of these four EP receptors capable of recognizing PGE 2 . The development of aberrant crypt foci and putative preneoplastic lesions in the colon was decreased in the EP4 knock-out mice (7). Blockade of EP4 production also mediated inhibition of NSCLC cell inva- siveness (10). Here, we reported that Fn only induced the expression of EP4, whereas it had no effect on other EP receptors. A similar lack of changes in the other EP receptors was observed when EP4 was knocked down by EP4 siRNA. This finding indicates that Fn selectively targets EP4 receptor subtypes in NSCLC cells. In view of the above, we focused on EP4 and explored the mechanisms involved.
First, we demonstrated that EP4 siRNA antagonized Fn-induced lung carcinoma cell growth, suggesting a direct role for EP4 in mediating this process in our system. This is consistent with data from others who reported that EP4 antisense oligonucleotides diminished EP4 protein expression and abolished the PGE 2 -stimulated production of cAMP and blocked the ability of PGE 2 to augment release of immunoreactive substance P and calcitonin gene-related peptide in sensory neurons (33). The EP4 antagonist, ONO-AE3-208, and the EP4 siRNA have been shown to inhibit extracellular matrix-induced metalloproteinase-9 expression in macrophages (34). More recently, EP4 antagonists inhibited breast cancer cell growth and reduced breast, lung, and colon cancer metastasis, suggesting that blockade of EP4 may be an alternative approach to the use of COX-2 inhibitors to prevent tumor metastasis (35,36).
In this study, we confirm that Fn increased EP4 gene expression in NSCLC cells, whereas collagen type 1 had no effect. This suggested that Fn may induce NSCLC growth not only by stimulation of PGE 2 but may also enhance this process by inducing the expression of EP4. The connection of Fn and EP4 expression has never been reported in lung cancer cells, although an antagonist of the PGE 2 receptor EP1 has been shown to decrease Fn synthesis in a rat diabetic nephropathy model (37). Also, PGE 2 -accelerated ProNectin F(TM) (a proteolytic fragment of Fn)-dependent adhesion was mediated through coop- FIGURE 5. Electrophoretic mobility shift assay to determine DNA binding of AP-2, Sp1, and C/EBP protein in the EP4 promoter in response to Fn. A-C, effect of Fn on C/EBP, Sp1, and AP-2 binding activities. Oligonucleotides containing the C/EBP (A), Sp1 (B), and AP-2 (C) sites were end-labeled with [␥-32 P]ATP and incubated with nuclear extracts (5 g) from H1838 cells treated with Fn or collagen type 1 (20 g/ml each) for an additional 24 h. D, ERK antisense blocks Fn-induced AP-2 binding activities. Top, cellular protein was isolated from H1838 cells transfected with control oligonucleotide or ERK antisense oligonucleotide (1 M each) for 24 h and then subjected to Western blot analysis for ERK1 and ERK2 protein. Bottom, oligonucleotides containing AP-2 sites were end-labeled with [␥-32 P]ATP and incubated with nuclear extracts (5 g) from H1838 cells transfected with 1 M ERK antisense or sense oligonucleotide for 24 h before exposing the cells to Fn (20 g/ml)-coated culture plates for an additional 24 h. E, the inhibitors of ERK1/2 and PI3K abrogated the Fn-induced AP-2 binding activities. Oligonucleotides containing AP-2 sites were end-labeled with [␥-32 P]ATP and incubated with nuclear extracts (5 g) from H1838 cells treated with PD98905 (25 M) and wortmannin (100 nM) for 1 h before exposing the cells to culture plates coated with Fn (20 g/ml) for an additional 24 h. F, anti-AP-2␣ antibody supershift. Oligonucleotides containing AP-2 sites were end-labeled with [␥-32 P]ATP and incubated with nuclear extracts (5 g) and AP-2␣, -␤, and -␦ antibodies (2 g/l each) for 24 h. For competition assays, a molar excess (100ϫ) of consensus Sp1 (Cold Sp1) or C/EBP (Cold C/EBP) or AP-2 (Cold AP-2) oligonucleotide were added to the binding reaction. Oligonucleotides containing a mutated Sp1 (Mut Sp1) or C/EBP (Mut C/EBP) or AP-2 (Mut AP-2) site that were end-labeled with [␥-32 P]ATP were used to confirm the binding specificity. Con, untreated control cells.
We previously demonstrated that Fn stimulated lung carcinoma cell growth through its receptor ␣5␤1, since anti-␣5␤1 antibodies eliminated the mitogenic response (4). Here, the Fn effect on EP4 protein levels was blocked by anti-␣5␤1 antibodies but not by anti-␣2␤1 antibodies, indicating that the integrin receptor ␣5␤1 mediated this regulation. In order to elucidate the mechanism(s) involved in Fn induction of EP4, we attempted to delineate the signaling pathways involved in induction of EP4 expression in lung carcinoma cells in response to Fn treatment. Data from our laboratory (4,21,24,39,40) and data of others (3,41) have demonstrated that adhesion to Fn activates several kinase signaling pathways, including ERK, PI3K, PKC, and PKA. We found that inhibitors of the PI3K kinase and ERK prevented Fn-induced EP4 protein expression, suggesting that the activation of these dual kinase signaling pathways is necessary for Fn-induced EP4 expression. We previously demonstrated that Fn up-regulated COX-2 expression through activation of the ERK signal pathway and that blockade of this signal abrogated Fn-induced COX-2 expression (4). These together suggested a strong connection between COX-2, EP4, and ERK signals. However, ERK played no role in the increase of EP4 expression induced by peroxisome proliferatoractivated receptor ␤/␦ activation (15), suggesting the existence of independent pathways that differ according to the stimulus. Fn activates MMP-9 via the ERK and PI3K/Akt signaling pathways in NSCLC and ovarian cancer cells (24,42,43). The inhibitor of PI3K, wortmannin, blocked the effect of Fn on stimulation of ERK phosphorylation, indicating cross-talk between the PI3K and ERK1/2 pathways in NSCLC cells. The cross-talk between these kinases has been reported in other cell systems as well (44,45). In contrast, our data indicated that mTOR, PKC, and PKA signaling pathways are not involved in the up-regulation of the EP4 gene induced by Fn, although the latter two kinases were involved in prostaglandin E receptor expression in other studies (46,47). This might suggest distinct effects, depending on the stimulant and cells studied.
EP4 has been shown to be regulated at the level of gene transcription in different cell types (21,47). We found that Fn, not collagen type 1, increased EP4 promoter activity. Furthermore, the region between Ϫ1555 and Ϫ992 was demonstrated to play Ϫ116) together with the AP-2 antisense or sense ODN (4 M) for 24 h, and then the cells were exposed to the culture plates coated with 20 g/ml Fn for an additional 24 h. D, effects of Fn and collagen type 1 on mutated EP4 promoter activities. H1838 cells were transfected with three EP4 mutated constructs (Ϫ1529, Ϫ1133, and Ϫ1000 bp with mutated AP-2 sites) for 24 h and then treated with 20 g/ml Fn or Collagen type 1 for additional 24 h. The ratio of firefly luciferase to Renilla luciferase activity was quantified as described under "Experimental Procedures." The bars represent the mean Ϯ S.D. of at least four independent experiments for each condition. *, significance as compared with controls. **, significance of combination treatment as compared with Fn alone (p Ͻ 0.05). Con, untreated control cells. Cellular protein was isolated from H1838 cells exposed to culture plates coated with Fn (20 g/ml) at the indicated time periods. Afterward, Western blot analysis was performed using polyclonal antibodies against AP-2␣, AP-2␤, and AP-2␦. B, dose-dependent effect of Fn on AP-2 protein expression. Cellular proteins were isolated from H1838 cells exposed to different concentrations of Fn for 24 h. Afterward, Western blot analysis was performed using polyclonal antibodies against AP-2␣, AP-2␤, and AP-2␦. C, effect of ERK inhibitor on Fn-induced AP-2␣ protein expression. Cellular protein was isolated from H1838 cells treated with PD98095 (25 M) for 2 h before exposing the cells to culture plates coated with Fn (20 g/ml) for an additional 24 h. D, effect of PI3K inhibitor on Fn-induced AP-2␣ protein expression. Cellular protein were isolated from H1838 cells treated with wortmannin (0.1 M) for 2 h before exposing the cells to culture plates coated with Fn (20 g/ml) for an additional 24 h. E, anti-␣5␤1 antibodies block the effect of Fn on AP-2␣ protein expression. Cellular protein was isolated from H1838 cells treated with anti-␣5␤1 antibody or anti-␣2␤1 (25 g/ml each) for 2 h before exposing the cells to culture plates coated with Fn (20 g/ml) for an additional 24 h. Afterward, Western blot analysis was performed to detect AP-2␣ protein levels. Actin served as internal control for normalization purposes. Con, untreated control cells.
a major role. The results showing that both PD98095 and wortmannin partially prevented Fn-stimulated EP4 promoter activity further suggested a role for PI3K and ERK kinase signaling pathways in mediating Fn up-regulation of EP4 gene expression. Several transcription factor binding sites within regions of the EP4 promoter have been characterized, including regulatory elements for AP-2, C/EBP, Sp1, and others (15,18,25). We showed that treatment of H1838 cells with Fn significantly increased protein binding activities of AP-2 in the EP4 promoter, whereas it had little effect on C/EBP and Sp1. This, together with the supershift assay results, indicated that AP-2␣ binding to the AP-2 site was necessary for the up-regulation of EP4 gene transcription in response to Fn. There are three AP-2 binding sites in this region. By mutations of each of these sites, we found that only one site (Ϫ1000 bp) was involved in the Fn-induced EP4 promoter activity. To our knowledge, a role for the AP-2 site in regulation of EP4 expression has never been reported. The transcription factor AP-2 regulates genes involved in a spectrum of important biological functions. Data obtained from different experimental models in vitro and in vivo indicate that AP-2 proteins function as important regulators of c-Myc targets in cell cycle progression and apoptosis (48). AP-2␣ overexpression leads to anchorage-independent growth and malignant transformation in vitro (49). It has therefore been suggested that AP-2 is involved in malignant transformation of breast cancer cells. Immunohistochemical analyses with AP-2␣and AP-2␥-specific reagents confirmed up-regulation of both proteins in human breast cancers (50). Other data suggested that AP-2␣ might be required for cell proliferation by suppression of genes inducing terminal differentiation, apoptosis, and growth retardation (51).
There are a few studies that explore the transcriptional regulation of EP4 and transcription factor interactions in its promoter region. Studies show that a GC-rich/Sp1 binding site located within the first 80 bases of the transcription start site in the EP4 promoter region is important in transcription initiation of the EP4 gene (24), and several negative, positive and lipopolysaccharide/ serum-responsive regions are located at different areas in the mouse EP4 promoter (16). We confirmed that AP-2 sites were involved in Fn-induced EP4 gene expression using mutated EP4 constructs.
Fn increased nuclear AP-2␣ protein levels, and inhibition of ERK signals prevented Fn-induced AP-2 expression. The connection between the AP-2 and ERK signal has been reported in other studies. For example, the ability of estradiol to increase AP-2 protein expression and AP-2 DNA binding activity was reversed by PD98059 (52). ERK activation is necessary for induction of the binding activities of AP-2 in T cells (53). Also, increased ERK signal has been reported in the absence of AP-2␣ in mouse epidermis (54). In addition, we found that blockade of AP-2 by AP-2 antisense approaches had no effect on ERK activ- FIGURE 8. The role of COX-2 signaling in mediating the effect of Fn on EP4 and AP-2 protein expression and inducing lung carcinoma cell growth. A, effect of blocking COX-2 signal on Fn-induced EP4 protein expression. Cellular protein was isolated from H1838 cells transfected with control or COX-2 siRNA (100 nM each) for 30 h before exposing the cells to culture plates coated with Fn for an additional 24 h and then subjected to Western blot analysis for COX-2, EP4, and actin proteins. B, effect of blocking COX-2 signal on Fn-induced AP-2␣ protein expression. Cellular protein was isolated from H1838 cells transfected with control or COX-2 siRNA (100 nM each) for 30 h before exposing the cells to culture plates coated with Fn (20 g/ml) for an additional 24 h and then subjecting them to Western blot analysis for COX-2, AP-2␣, and actin protein levels. C, effect of PGE 2 on AP-2␣ protein expression. Cellular protein was isolated from H1838 cells treated with or without dmPGE 2 (1 M) for 4 h before exposing the cells to culture plates coated with Fn (20 g/ml) for an additional 24 h. Afterward, Western blot analysis was performed to detect AP-2␣ protein levels. D, effect of PGE 2 on EP4 protein expression. Cellular protein was isolated from H1838 cells treated with or without dmPGE 2 (1 M) for 4 h before exposing the cells to culture plates coated with Fn (20 g/ml) for an additional 24 h. Afterward, Western blot analysis was performed to detect EP4 protein levels. Actin served as internal control for normalization purposes. E and F, effect of blocking EP4 signal on PGE 2 -induced lung carcinoma cell growth. E, H1838 cells were transfected with control or EP4 siRNA pool for 30 h before exposing the cells to dmPGE 2 (0.1 M) and incubation with 1 Ci/ml [methyl-3 H]thymidine for an additional 24 h as indicated. Data are expressed as mean Ϯ S.D. of at least three independent experiments. F, H1838 cells were transfected with control or EP4 siRNA pool for 30 h before exposing the cells to dmPGE 2 (0.1 M) for an additional 48 h, Afterward, viable cell numbers were determined by the CellTiter-Glo luminescent cell viability assay. All data are presented as means Ϯ S.D. *, significant difference from control. **, significance of combination treatment as compared with dmPGE 2 alone (p Ͻ 0.05). Con, untreated control cells.

Fibronectin Stimulates EP4 in Lung Cancer
ity in the presence or absence of PGE 2 (not shown), suggesting that ERK was not a downstream signal of AP-2. Furthermore, we showed that inhibition of PI3K and blockade of ␣5␤1 integrin signals abrogated Fn-induced AP-2␣ protein expression, indicating that multiple signals were involved in this process. By showing that AP-2 antisense oligonucleotides abolished Fn-induced cell growth, our data suggest that induction of AP-2␣ by Fn is at least partly responsible for its effect on NSCLC proliferation.
We previously demonstrated that Fn increased COX-2 gene expression and stimulation of PGE 2 production (4). The COX-2 siRNA abrogated Fn-induced EP4 protein expression, further indicating that COX-2 signals mediated the effect of Fn on EP4 expression. We showed that PGE 2 stimulated EP4 expression in NSCLC cells (15). Since COX-2 expression is involved in PGE 2 production, the effect of Fn on EP4 could be solely mediated through stimulation of PGE 2 production. In support of this, the literature contains data that link expression of COX-2 and AP-2 signaling (55, 56). PGE 2 not only stimulated the base line but also enhanced Fn-induced AP-2␣ protein expression and cell proliferation, suggesting that stimulation of PGE 2 by Fn might account, at least partially, for the production of AP-2␣ that resulted in EP4 expression and, subsequently, increased cell proliferation. PGE 2 has been shown to enhance the binding activity of AP-2 in different gene promoters in several studies (57,58). dmPGE 2 up-regulated expression of the receptor subtype EP4, suggesting that PGE 2 might combine with EP4 receptors in autocrine or paracrine modes. Thus, blockade of the EP4 gene by siRNA approaches diminished the stimulatory effect of PGE 2 on lung carcinoma cell growth (this study). The dose of PGE 2 used in this study was based on our previous results showing Fn-induced PGE 2 production in NSCLC cells in vitro (4). Similar or even higher doses of exogenous PGE 2 have been shown to suppress both Th1-and Th2-polarized antigen-specific human T-cell responses and to reduce radiation-induced apoptosis in human colon cancer cells without toxicity (59,60). Previously, we reported that PGE 2 enhanced NSCLC cell growth in the presence of the EP2 agonist, Butaprost, suggesting that EP2 also plays a role in mediating NSCLC cell growth (17). Whether EP2 is directly involved in the effect of PGE 2 on NSCLC cell proliferation and how this pathway is affected by Fn needs to be determined.
In summary, our studies show that Fn stimulates human lung carcinoma cell proliferation through the PGE 2 receptor subtype EP4. This effect is enhanced by Fn-induced EP4 expression. Control of EP4 gene expression by Fn is dependent on ␣5␤1 integrin-mediated signals that include activation of PI3K/ Akt, ERK, and COX-2. These signals stimulate PGE 2 production, which induces the transcription factor AP-2␣ and stimulates AP-2␣ interactions with critical DNA regions within the EP4 gene promoter (Ϫ1555 to Ϫ992 bp). Thus, Fn, by increased PGE 2 production and EP4 gene expression, induces mitogenic signals in NSCLC cells (Fig. 9). This study reveals a novel molecular mechanism for Fn regulation of human lung carcinoma cell growth and provides further evidence for the role of E prostanoid receptors in lung carcinoma biology.