The prostaglandin receptor EP2 activates multiple signaling pathways and β-arrestin1 complex formation during mouse skin papilloma development

Abstract

Prostaglandin E₂ (PGE₂) is elevated in many tumor types, but PGE₂'s contributions to tumor growth are largely unknown. To investigate PGE₂'s roles, the contributions of one of its receptors, EP2, were studied using the mouse skin initiation/promotion model. Initial studies indicated that protein kinase A (PKA), epidermal growth factor receptor (EGFR) and several effectors—cyclic adenosine 3′,5′-monophosphate response element-binding protein (CREB), H-Ras, Src, protein kinase B (AKT) and extracellular signal-regulated kinase (ERK)1/2—were activated in 12-O-tetradecanoylphorbol-13-acetate (TPA)-promoted papillomas and that PKA and EGFR inhibition (H89 and AG1478, respectively) decreased papilloma formation. EP2's contributions to the activation of these pathways and papilloma development were determined by inhibiting endogenous TPA-induced PGE₂ production with indomethacin (Indo) and concomitantly treating with the EP2 agonist, CAY10399 (CAY). CAY treatment restored papilloma formation in TPA/Indo-treated mice and increased cyclic adenosine 3′,5′-monophosphate and PKA activation as measured by p-CREB formation. CAY treatment also increased EGFR and Src activation and their inhibition by AG1478 and PP2 indicated that Src was upstream of EGFR. CAY also increased H-Ras, ERK1/2 and AKT activation, and AG1478 decreased their activation indicating EGFR being upstream. Supporting EP2's contribution, EP2−/− mice exhibited 65% fewer papillomas and reduced Src, EGFR, H-Ras, AKT and ERK1/2 activation. G protein-coupled receptor (GPCR) activation of EGFR has been reported to involve Src's activation via a GPCR–β-arrestin–Src complex. Indeed, immunoprecipitation of β-arrestin1 or p-Src indicated the presence of an EP2–β-arrestin1–p-Src complex in papillomas. The data indicated that EP2 contributed to tumor formation via activation of PKA and EGFR and that EP2 formed a complex with β-arrestin1 and Src that contributed to signaling and/or EP2 desensitization.

Introduction

Elevated prostaglandin (PG) levels are commonly observed in a variety of human and animal tumors. The cyclooxygenases (COX-1 and COX-2) are key enzymes in the generation of PGs from arachidonic acid; COX-2, the inducible isoform, is elevated in many tumor types (1). The major PG formed in most tumors is prostaglandin E₂ (PGE₂) that manifests its biological activities by binding to the G protein-coupled receptors (GPCRs): EP1, EP2, EP3 and EP4 (2). Studies in a variety of mouse cancer models using receptor-deficient mice and/or receptor antagonists have indicated the roles for EP1, EP2, EP3 or EP4 in colon, breast, lung, intestine and skin tumor development (3–8). However, the signaling pathways activated in vivo by the individual PGE₂ receptors and their contributions to tumor development have received little attention.

In GPCR-mediated signaling, the binding of the ligand to the receptor causes the activation and disassociation of the heterotrimeric G proteins, and the activated G proteins then regulate the activities of various effectors (9). For PGE₂-stimulated EP-mediated signaling, EP1 caused the activation of phospholipase C (2). EP2 and EP4, which have been extensively studied by Regan et al. (10–12), caused the activation of adenylate cyclase (AC), protein kinase A (PKA) and phosphatidylinositol 3-kinase/protein kinase B (AKT), and depending on the splice variant, EP3 increased or decreased AC activity (2). Early on it was thought that GPCR activity was regulated by receptor internalization/desensitization following the formation of a complex with β-arrestin1 or 2 (9,13). However, more recent studies have indicated that GPCR–β-arrestin complexes have more biological functions than desensitization, as the complex can also contribute to receptor-mediated signaling, including the activation of Src, epidermal growth factor receptor (EGFR) and extracellular signal-regulated kinase (ERK)1/2 (9,13,14). Indeed, a recent in vitro study demonstrated that PGE₂ stimulation of EP4 caused the formation of an EP4–β-arrestin1–Src complex that resulted in Src activation and the transactivation of EGFR (15). Thus, ligand-activated GPCRs have the capabilities of activating signaling pathways by G protein-dependent as well as G protein-independent mechanisms.

The mouse skin initiation/promotion [7,12-dimethylbenz(a)anthracene (DMBA)/12-O-tetradecanoylphorbol-13-acetate (TPA)] model has proven to be a useful model to study tumor development (16). In addition, this model has been used to demonstrate that PGs play an important role in skin tumor formation (17). Both our previous work using COX-1- and COX-2-deficient mice and the work of others using non-steroidal anti-inflammatory drugs (NSAIDs) indicated that papilloma formation was reduced in initiated/promoted mice (18,19). Additionally, several signaling effectors, including EGFR, Src, Ras, AKT and ERK1/2, have been shown to be involved in skin tumor formation in the initiation/promotion model (16,20,21). However, it is unclear whether PGs and their receptors are involved in the activation of these pathways/effectors during papilloma development.

While most previous studies investigating GPCR-mediated signaling have utilized cell culture systems, in the present study, we utilized the mouse skin initiation/promotion model to elucidate the contributions of EP2 signaling to skin tumor development. EP2 was investigated because preliminary studies utilizing EP1–4 agonists and/or antagonist- and EP2-deficient mice indicated that EP2 played an important role in tumor formation in initiated/promoted mice. Furthermore, based on our observation that TPA increased COX-2/PGE₂ levels in skin (18) and studies demonstrating that PGE₂/EP2 activated PKA and EGFR signaling in cultured cells (2,22–24), we hypothesized that EP2-mediated signaling contributed to TPA-induced effects. Therefore, initial studies were conducted to demonstrate TPA's ability to activate PKA and EGFR signaling in mouse skin and papillomas and showed that PKA and EGFR inhibition (H89 and AG1478, respectively) could reduce papilloma formation. EP2's involvement in the activation of PKA and EGFR pathways was then determined using two approaches. First, the effects of the EP2-selective agonist, Cayman10399 (CAY) (6,25), on papilloma formation and signaling in TPA-promoted skin and papillomas were studied when endogenous TPA-induced PG production was inhibited by indomethacin (Indo). Second, papilloma formation and signaling differences between initiated/promoted EP2-deficient and wild-type (WT) mice were compared. The data indicated that EP2-mediated signaling increased the level of cyclic adenosine 3′,5′-monophosphate (cAMP) and PKA activation, as well as the activation of EGFR, Src, Ras, ERK and AKT and that the levels of these effectors were elevated in papillomas compared with surrounding skin. The data further indicated that EP2 formed a complex with β-arrestin1 and p-Src that could contribute to G protein-independent signaling and/or EP2 desensitization.

Materials and methods

Chemicals

DMBA and the actin antibody were purchased from Sigma–Aldrich Corp. (St Louis, MO) and TPA was purchased from Axxora LLC (San Diego, CA). PGE₂, CAY and the anti-EP2 antibody were obtained from Cayman Chemical (Ann Arbor, MI). H89, PP2 and AG1478 were obtained from EMD Chemicals/Calbiochem (San Diego, CA). Antibodies for AKT, p-AKT (Ser473), ERK1/2, p-ERK1/2 (Thr202/Tyr204), Src and p-Src (Tyr416), EGFR and p-EGFR (Tyr845) and p-CREB (Ser133) were obtained from Cell Signaling Technology (Danvers, MA), and the H-Ras antibody was purchased from BD Biosciences (San Jose, CA). The antibody for β-tubulin was purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and the β-arrestin1 antibody from Abcam (Cambridge, MA).

Animals and animal treatments

CD-1 female mice, six weeks of age were purchased from Charles River Laboratories (Raleigh, NC) and housed in the National Institute of Environmental Health Sciences animal facilities according to the Association for the Assessment and Accreditation of Laboratory Animal Care Guidelines. EP2−/− mice (26) were obtained from Dr Richard Breyer (Vanderbilt University, Nashville, TN) and maintained on a C57BL/6 background. Animal studies were approved by the National Institute of Environmental Health Sciences Animal Care and Use Committee. Food and water were provided ad libitum.

For the initiation/promotion studies, the dorsal skin of the mice was shaved and mice in the telogen phase of hair cycle were used. In all studies, mice were initiated with a single application of DMBA (50 μg in 200 μl acetone) and 1 week later treated twice a week with TPA (4 μg per treatment). Conditions for treatment with TPA, TPA/Indo or TPA/Indo plus CAY are indicated in the legends for the individual experiments. Conditions for studying the effects of the inhibitors, H89, AG1478 or PP2 on signaling or papilloma formation are also indicated in the appropriate figure legend.

PGE₂ determination

Papillomas and surrounding skin samples were excised, minced into small pieces and immediately frozen in liquid nitrogen. A piece of tissue weighing between 0.10 and 0.15 g was homogenized in cold phosphate-buffered saline containing 50 μM Indo. After centrifugation, appropriate amount of the supernatant from each sample was used for PGE₂ assay by using a radioimmunoassay kit (GE Healthcare Life Sciences, Piscataway, NJ) according to the manufacturer's instructions.

Western analysis and immunoprecipitation

Papillomas and surrounding skin samples from each of the three individual initiated/promoted mice were isolated and individually prepared for western analysis as described below. In Figure 1A, western data for surrounding skin and a papilloma from each of three individual mice are shown. Because variations existed between samples from individual mice in Figure 1A, for studies to determine the role of EP2-mediated signaling (Figures 3–5), skin tissue from three individual mice and two to three papillomas (∼4 mm diameter) from each of three individual mice were combined (six to nine papillomas total) for western analysis. The tissue samples were homogenized in cell lysis buffer (Cell Signaling Technology). Protein (50–80 μg) from each sample was electrophoresed and electroblotted to polyvinylidene difluoride membranes and the membranes were incubated overnight at 4°C with 1:500–1000 dilution of the specific antibody. Equal lane loading was assessed using β-tubulin or actin. For protein detection, the blots were incubated with horseradish peroxidase-conjugated secondary anti-rabbit or anti-mouse antibodies (Cell Signaling Technology). For immunoprecipitation, skin and papilloma lysates containing 200 μg proteins were incubated with the primary antibody and 20 μl of appropriate suspended protein A/G-agarose (Santa Cruz Biotechnology) as per the supplier's instructions. The immunoprecipitates were collected by centrifugation at 4°C, washed with cell lysis buffer and dissolved in loading buffer for western analysis. Each immunoprecipitation and western analysis was repeated with skin/papillomas from a separate group of mice.

Fig. 1.

Comparison of TPA-induced PKA and EGFR signaling in surrounding skin and papillomas and the effects of their inhibition on papilloma formation. (A) Differences in p-CREB (Ser133), EGFR, p-EGFR (Tyr845), H-Ras, p-Src (Tyr416), p-AKT (Ser473) and p-ERK1/2 (Thr202/Tyr204) levels in papillomas and surrounding skin as determined by western blotting. Actin served as a control for protein loading and membrane transfer and each lane represents an individual mouse. Intensities of the bands were determined by densitometry and the relative intensities from three individual skins or papillomas are presented as mean value ± SD, *P < 0.05 versus surrounding skin. The data are representative of two independent experiments. (B) Inhibition of PKA and EGFR suppresses papilloma formation in initiated/promoted mice. Fifteen mice per group were topically treated twice per week with TPA, TPA/H89 (250 μg) or TPA/AG1478 (30 μg) for 15 weeks. The inhibitors in 200 μl of acetone were applied 30 min before the TPA in 200 μl acetone. Data are expressed as the mean ± SD (n = 15) of the number of papilloma per mouse in each group. *P < 0.05 versus TPA-treated mice.

Fig. 3.

CAY increased cAMP/PKA and Src/EGFR activation in papillomas and surrounding skin. In (B–E), skin or two to three papillomas from three individual mice were combined (six to nine papillomas total) for analysis, and each experiment was repeated. (A) CAY increased cAMP levels in papillomas from TPA/Indo-treated mice. cAMP levels in papillomas and surrounding skin from mice treated twice per week for 15 weeks with TPA, TPA/Indo or TPA/Indo/CAY (5 μg per treatment) from Figure 2B above were determined. Samples were obtained 2 h after last treatment. *P < 0.05 versus surrounding skin; **P < 0.05 versus TPA-treated papillomas; ***P < 0.05 versus TPA/Indo-treated papillomas. (B) CAY restored p-CREB (Ser133) levels. The p-CREB (Ser133) levels and loading control, β-tubulin, were determined from papillomas (P) and surrounding skin (S) of mice treated as indicated. Fifty micrograms protein was loaded for western blotting from lysates prepared 24 h following the last treatment. (C) PKA inhibition decreased p-CREB levels in papillomas. Papillomas were generated by TPA/Indo/CAY promotion for 15 weeks and the treatment continued for two additional weeks with and without H89 (250 μg per treatment). Lysates were prepared from papillomas 24 h following the last treatment and p-CREB (Ser133), p-EGFR, p-ERK (Thr202/Tyr204) and p-AKT (Ser473) were detected by western blotting (50 μg protein). (D) CAY treatment increased p-EGFR and p-Src in TPA/Indo-treated papillomas. Initiated CD-1 mice (15 mice per group) were treated for 15 weeks with TPA, TPA/Indo or TPA/Indo/CAY (5 μg). Lysates were prepared 24 h following the last treatment from papillomas (P) and surrounding skin (S) and p-Src and p-EGFR detected by western blotting (75 μg protein). (E) Src kinase inhibition decreased p-EGFR levels in papillomas. Papillomas were generated by TPA/Indo/CAY (5 μg) treatment for 15 weeks and the treatment continued for two additional weeks with or without AG1478 (30 μg per treatment) or PP2 (30 μg per treatment). Lysates were prepared from papillomas 24 h following the last treatment and the levels of p-Src (Tyr416), p-EGFR (Tyr845) and the loading control, β-tubulin, detected by western blotting (75 μg protein).

Fig. 4.

EP2 contributed to the activation of H-Ras, ERK and AKT in papillomas (P) and surrounding skin (S). In (A–C), skin or two to three papillomas from three individual mice were combined (six to nine papillomas total) for analysis, and each experiment was repeated. (A) CAY treatment increased Ras, ERK1/2 and AKT activation in TPA/Indo-treated papillomas and surrounding skin. Lysates were prepared from skin and papilloma treated twice per week for 15 weeks with TPA, TPA/Indo or TPA/Indo/CAY and Ras-GTP (pull-down), p-AKT and p-ERK1/2 detected by western blotting (50 μg protein). (B) EP2 deficiency decreased Ras, ERK and AKT activation in papillomas. Western blotting (50 μg protein) of lysates was prepared from papillomas and surrounding skin from initiated/promoted WT and EP2−/− mice. (C) EGFR inhibition decreased Ras, ERK1/2 and AKT activation. Papilloma-bearing mice were generated by TPA/Indo/CAY (5 μg) for 15 weeks and the treatment continued for two additional weeks with or without AG1478 (30 μg per treatment). Lysates were prepared from papillomas 24 h following the last treatment and Ras-GTP (pull-down), p-AKT and p-ERK1/2 detected by western blotting (60 μg protein).

Fig. 5.

EP2 formed a complex with β-arrestin1 and p-Src. In (A–D), skin or two to three papillomas from three individual mice were combined (six to nine papillomas total) for analysis, and each experiment was repeated. (A) EP2 immunoprecipitated with β-arrestin1 from WT but not EP2−/− mice. EP2 immunoprecipitates were prepared from lysates (200 μg protein) of papillomas (P) and surrounding skin (S) from WT and EP2−/− mice and β-arrestin1 was detected by western blotting. (B) CAY increased EP2 and β-arrestin1 complex formation in p-Src immunoprecipitates of papillomas. Lysates were prepared from papillomas and surrounding skin from CD-1 mice treated twice per week for 15 weeks with TPA/Indo or TPA/Indo/CAY (5 μg). Twenty-four hours following the final treatment, p-Src immunoprecipitates (200 μg protein) were prepared and EP2, β-arrestin1 and p-Src detected by western blotting. (C) EP2, β-arrestin1 and p-Src complex formation was elevated in TPA-treated WT compared with EP2−/− papillomas. p-Src immunoprecipitates were prepared from lysates (200 μg protein) of papilloma and surrounding skin from WT and EP2−/− mice and the levels of EP2, β-arrestin1, p-Src and p-EGFR detected by western blotting. (D) p-EGFR formed a complex with p-Src independent of the EP2–β-arrestin1–p-Src complex. β-Arrestin1 immunoprecipitates were prepared from lysates (200 μg protein) of papillomas from WT mice and the levels of p-Src, EP2, p-EGFR and β-arrestin1 determined by western blotting.

cAMP immunoassays

cAMP measurement was carried out using a kit from BioVision (Mountain View, CA). Briefly, the protein fraction (150 μg) from the dorsal skin of mice was diluted with 0.1 N HCl. The samples were added to protein A-coated 96-well plates and incubations carried out with a cAMP antibody for 1 h at room temperature. cAMP–horseradish peroxidase was added to each well, incubated for 1 h at room temperature, washed with assay buffer and horseradish peroxidase developer added to each well. Reactions were terminated by the addition of 1 N HCl and the optical density was determined immediately at 450 nm. Calculations were based on a standard curve for each experiment.

Ras-GTP assays

Activated Ras levels were determined from tissue lysates using the Ras Activation Kit from Assay Designs (Ann Arbor, MI) according to the manufacturer's instructions. Briefly, lysates (250 μg protein) from surrounding skin and papillomas were affinity precipitated using 80 μg of recombinant glutathione S-transferase–Raf1 Ras-binding domain fusion protein for 1 h at 4°C with gentle rocking. The precipitates were washed three times with lysis/binding/wash buffer and eluted with 50 μl of 2× sodium dodecyl sulfate–polyacrylamide gel electrophoresis sample buffer containing β-mercaptoethanol. The proteins were separated on a 12% sodium dodecyl sulfate–polyacrylamide gel and then immunoblotted with a H-Ras antibody.

Statistical methods

Statistical analyses were performed using Student's t-test and a P value of <0.05 was considered statistically significant.

Results

TPA increases PKA and EGFR signaling in papillomas and their inhibition decreases papilloma formation

In a previous study, we demonstrated that TPA promotion increased COX-2 and PGE₂ levels to a greater extent in papillomas compared with surrounding skin (18) and hypothesized that TPA's tumor-promoting activity, in part, involved PGE₂ activation of EP2. Because PGE₂/EP2 has been shown to activate PKA and EGFR (2,6,22–24), we determined if the PKA- and EGFR-signaling pathways were elevated in TPA-promoted papillomas. Figure 1A shows that p-CREB, a transcription factor activated by PKA, was increased ∼3-fold in TPA-promoted papillomas compared with surrounding skin. EGFR, a receptor involved in papilloma formation (16,20,27), and p-EGFR levels were elevated ∼4- and 8-fold, respectively, in papillomas compared with surrounding skin. As expected (28,29), total H-Ras was elevated ∼5-fold in papillomas, and the activated form of Src, an oncogene known to contribute to papilloma formation (21), was elevated ∼5-fold. Phosphorylated AKT and ERK1/2, effectors known to contribute to papilloma formation (30–33), were elevated ∼4- and 8-fold, respectively, in papillomas compared with surrounding skin. Total cyclic adenosine 3′,5′-monophosphate response element-binding protein (CREB), Src, AKT and ERK1/2 levels were unchanged in TPA-promoted papillomas compared with surrounding skin (data not shown). The data in Figure 1B further illustrate the importance of PKA and EGFR signaling to TPA-promoted papilloma development as their inhibition by H89 and AG1478, respectively, decreased papilloma formation by ∼50%. Thus, the data indicated that EP2 contributed to TPA's activation of PKA and EGFR and that activated PKA and EGFR contributed to papilloma formation. In the studies below, EP2’s contribution to TPA-promoted papilloma formation is described.

Activation of EP2 contributes to TPA-promoted papilloma development

To determine if EP2 contributed to TPA-promoted papilloma development, DMBA-initiated mice were treated for 15 weeks with TPA, TPA plus Indo (to block endogenous PG production) or TPA/Indo plus PGE₂ or the EP2 agonist, CAY. Figure 2A shows that topically applied Indo decreased PGE₂ production ∼70% in acetone-treated skin and ∼80% in TPA-treated skin. Figure 2B shows that Indo treatment also reduced the number of papillomas on TPA-promoted mice from ∼30 papillomas per mouse to ∼8 papillomas per mouse and that exogenous PGE₂ or CAY treatment restored papilloma formation in TPA/Indo-treated mice. Figure 2C shows that CAY treatment of TPA/Indo-treated mice produced a dose-dependent increase in papilloma formation. Furthermore, Figure 2D illustrates that while Indo treatment reduced papilloma size in TPA-treated mice, papilloma size was restored by CAY treatment. The observation that initiated/promoted EP2−/− mice developed ∼65% fewer tumors than WT mice (Figure 2E), in agreement with the report by Sung et al. (5), confirmed a role for EP2 in papilloma development.

Fig. 2.

CAY increased papilloma formation and epidermal hyperplasia. (A) PGE₂ levels in skin were determined, using skin samples from CD-1 mice receiving one of the following treatments twice per week for 4 weeks: acetone, acetone/Indo, TPA (4.0 μg per treatment) or TPA/Indo. *P < 0.05 versus acetone-treated group; **P < 0.01 versus TPA-treated group. (B) CAY and PGE₂ increased papilloma formation in TPA/Indo-treated mice. DMBA-initiated CD-1 mice (15 mice per group) were treated twice per week for 15 weeks with TPA, TPA plus Indo, TPA/Indo plus PGE₂ or CAY (both 5 μg per treatment). The data are expressed as the number of papillomas per mouse (mean ± SD) at 15 weeks. *P < 0.01 versus TPA-treated group; **P < 0.01 versus TPA/Indo-treated group. (C) CAY induced a dose response in papillomas in TPA/Indo-treated mice. DMBA-initiated mice (15 mice per group) were treated twice per week for 15 weeks with TPA, TPA/Indo or TPA/Indo/CAY with the indicated doses of CAY. *P < 0.01 versus TPA-treated group, **P < 0.05 versus TPA/Indo-treated group. (D) Papillomas size distribution in TPA, TPA/Indo and TPA/Indo/CAY-treated mice [from (B) above]. (E) EP2 deficiency decreased papilloma numbers in initiated/promoted mice. WT (black bars) and EP2−/− mice (striped bars) were treated with TPA twice per week and the mean number of papillomas per mouse was determined weekly. *P < 0.05 versus WT group. (F) Hematoxylin and eosin-stained skin sections from CD-1 mice treated twice per week for 4 weeks with TPA. The scale bar is 10 μm. (G) TPA/Indo or (H) TPA/Indo/CAY (5 μg per treatment). (I) Bromodeoxyuridine (BrdU)-labeled cells in (F–H) (above). The mice were injected with BrdU for 2 h at 22 h after last TPA treatment. The skin sections were immunostained with an antibody against BrdU. BrdU-positive cells were counted in three random skin sections from each of five mice. Data are expressed as the mean ± SD (n = 3) of the number of stained cells per 100 basal cells in each group. *P < 0.05 versus TPA-treated group; **P < 0.01 versus TPA/Indo-treated group.

Because the EP2 agonist increased papilloma numbers and size in TPA/Indo-treated mice (Figures 2C and D), the effect of CAY on epidermal thickness was determined. Figure 2F and G shows that TPA-induced epidermal hyperplasia was reduced by Indo treatment but that CAY treatment restored the hyperplastic response (Figure 2H). To account for CAY's hyperplastic effects, it was observed that CAY treatment increased bromodeoxyuridine incorporation in TPA/Indo-treated mice by ∼100% (Figure 2I). Because the data indicated that EP2 played a significant role in epidermal cell replication and papilloma development, EP2-mediated signaling in papillomas and surrounding skin was investigated.

EP2 signaling increases cAMP, PKA activation and EGFR activation

Because the data in Figure 1B indicated that PKA signaling had a role in TPA-promoted papilloma formation, the possibility that EP2 contributed to the activation of PKA in skin and papillomas was investigated. Cell culture studies had indicated that EP2 signaling involved G_αs activation of AC, which increased cAMP levels and the activation of PKA (10). Figure 3A shows that cAMP levels were increased in TPA-promoted papillomas compared with surrounding skin and that Indo treatment decreased cAMP levels in papillomas. The data further show that CAY treatment restored the reduction in cAMP levels caused by Indo. In addition, Figure 3B shows that CAY also increased PKA activation in TPA/Indo-treated papillomas when measured by increased phosphorylation of CREB at a PKA-specific site (Ser133) (34). To confirm that PKA mediated EP2’s activation of p-CREB, treatment with the PKA inhibitor, H89, reduced p-CREB levels in TPA/Indo/CAY-promoted papillomas (Figure 3C). However, Figure 3C indicates that H89 at the dose used (30 μg per treatment) did not inhibit EGFR, ERK1/2 or AKT activation in CAY-treated papillomas. Total CREB, EGFR, ERK1/2 and AKT were not affected by H89 treatment (data not shown). Thus, the data indicate that EP2 contributed to the activation of PKA but that the activation of EGFR, ERK1/2 and AKT occurred by PKA-independent pathways.

EGFR is known to have an important role in the development of various tumor types. Of relevance to skin tumor development, previous studies demonstrated that EGFR deficiency reduced papilloma size when H-Ras-transformed keratinocytes were grafted onto nude mice (27) and the data in Figure 1B indicate that EGFR inhibition reduced mouse skin papilloma formation. Furthermore, recent studies using cultured cells have indicated that PGE₂ activated EGFR via Src-mediated pathways (22,35). Figure 3D shows that the levels of p-EGFR, phosphorylated at a Src-specific site (Tyr845), and p-Src (Tyr416) in TPA/Indo-treated papillomas were increased by CAY treatment. To determine if p-Src mediated EP2’s activation of EGFR or vice versa, papilloma-bearing mice were treated topically with the EGFR inhibitor, AG1478, or the Src inhibitor, PP2. The data in Figure 3E show that the level of p-Src in papillomas was not significantly decreased by AG1478 treatment, whereas p-EGFR levels were significantly decreased by PP2 treatment. The observation that PP2 partially inhibited Src phosphorylation suggests that Src phosphorylation occurred both by Src autophosphorylation (PP2 inhibitable) as well as by other kinases (PP2 independent). Thus, the data indicate that EP2 contributed to the activation of Src and EGFR in papillomas and that p-Src contributes to the activation of EGFR.

EP2 influences the activation of H-Ras, ERK1/2 and AKT

Mutation of H-Ras and increased H-Ras levels are hallmarks of DMBA-initiated/TPA-promoted skin papillomas (16). In the present study, DNA sequencing indicated that nine of nine papillomas contained the expected codon 61 H-Ras mutation (data not shown). Figure 4A shows that activated H-Ras levels were increased in papillomas from TPA-treated mice compared with surrounding skin and that Indo co-treatment decreased activated H-Ras levels. Figure 4A further shows that CAY treatment increased the levels of both total and activated H-Ras in papillomas and surrounding skin of Indo-treated mice. In support of EP2 contributing to the regulation of total and activated H-Ras levels, Figure 4B shows that the levels of total and activated Ras were reduced in papillomas from EP2−/− mice compared with WT mice.

The data in Figures 1A and 4A indicate that the levels of p-ERK1/2 (Thr202/Tyr204) and p-AKT (Ser473) were elevated in papillomas from TPA-treated mice, and Figure 4A shows that Indo treatment significantly reduced their activation. The data in Figure 4A further show that CAY increased both ERK1/2 and AKT activation in papillomas of Indo-treated mice. Furthermore, ERK1/2 and AKT activation were reduced in skin and papillomas from TPA-treated EP2−/− compared with WT mice (Figure 4B). Thus, the data indicated that CAY activation of EP2 increased the levels of activated H-Ras as well as activated ERK1/2 and AKT.

Because EP2 contributed to the activation EGFR (Figure 3D), and EGFR can lead to the activation of H-Ras, ERK1/2 and AKT; the contribution of EP2-mediated activation of EGFR in the activation of H-Ras, ERK1/2 and AKT was investigated. Figure 4C indicates that AG1478 inhibition of EGFR in TPA/Indo/CAY-promoted papillomas significantly decreased the activation of H-Ras, ERK1/2 and AKT, suggesting that the activation of these effectors was, in part, downstream of EP2-mediated activation of EGFR.

EP2 forms a complex with β-arrestin1 and p-Src

Because p-Src appeared to be an intermediate by which EP2 contributed to the activation of EGFR (Figure 3E), a possible mechanism to account for EP2-mediated Src activation was investigated. It has been reported that GPCRs can be internalized by forming a complex with β-arrestin1 and/or 2 and that GPCR–β-arrestin complexes can complex with Src leading to the activation of Src and other signaling pathways (13,14). Figure 5A indicates that β-arrestin1 was detected by western blotting of EP2 immunoprecipitates from skin and papillomas of WT but not EP2−/− mice and indicates that EP2 formed a complex with β-arrestin1. To determine if the EP2–β-arrestin1 complex also involved p-Src, p-Src immunoprecipitates of skin and papilloma homogenates were analyzed for the presence of EP2 and β-arrestin1. Figure 5B shows that following p-Src immunoprecipitation, both EP2 and β-arrestin1 were detected by western blotting and that the level of the EP2–p-Src–β-arrestin1 complex was higher in papillomas from TPA/Indo/CAY-treated mice than TPA/Indo-treated mice. Furthermore, Figure 5C indicates that both β-arrestin1 and p-Src were present, but at lower levels, in p-Src immunoprecipitates of papillomas from EP2−/− mice compared with WT mice. These data indicate that in TPA-treated papillomas, EP2 activation contributes to β-arrestin1–p-Src complex formation (Figure 5C).

The presence of some immunoprecipitable β-arrestin1 and p-Src in papillomas of EP2−/− mice (Figure 5C) suggests that GPCRs, in addition to EP2, also form a complex with β-arrestin1 and p-Src. Figure 5C further indicates that p-Src can form a complex with p-EGFR and that the level of p-EGFR is greater in papillomas of WT than EP2−/− mice. Thus, these data are in agreement with Src being activated via an EP2–β-arrestin1 complex and that p-Src then mediates the transactivation of EGFR. However, the data do not indicate whether p-Src dissociates from the EP2–β-arrestin1 complex before activating EGFR. To address this issue, β-arrestin1 immunoprecipitates of papillomas from WT mice were utilized. The data in Figure 5D show that while EP2, p-Src and β-arrestin1 were detected in β-arrestin1 immunoprecipitates, p-EGFR was not detected. Therefore, the data suggest that p-Src disassociates from the EP2–β-arrestin1 complex before interacting with and activating EGFR.

Discussion

We previously reported that the genetic deficiency of COX-1 or COX-2 decreased papilloma formation when the initiation/promotion mouse skin protocol was used (18). Because PGE₂ was the major PG formed in skin, we hypothesized that diminished PGE₂ levels and therefore decreased PGE₂-induced signaling in the COX-deficient mice were responsible for decreased tumor formation. PGE₂ manifests its activity via four GPCRs, EP1, EP2, EP3 and EP4 (2); however, in the present study, we focused on the contribution of EP2 to mouse skin papilloma formation. Previous studies have reported that the levels of EP2 expression influenced mouse skin tumor formation in the initiation/promotion model (5,36) and on ultraviolet exposure (37,38). Using the initiation/promotion model, EP2 deficiency (5) was shown to decrease and EP2 overexpression (36) was shown to increase papilloma formation. However, in these studies, TPA-induced endogenous PGE₂ was the likely activator of EP2. Indeed, exogenous PGE₂ has been reported to activate the PKA and EGFR pathways in mouse keratinocytes and naive mouse skin, although a specific receptor was not identified (23). In addition, PKA and EGFR have been reported to be activated in vitro via EP2 in squamous cell carcinoma lines (24). In the present report, we extend these studies by utilizing an EP2-selective agonist and EP2-deficient mice to identify EP2-mediated signaling pathways activated during tumor development in the mouse skin initiation/promotion model. In addition, in contrast to previous studies (39,40), we report that EP2 formed a complex with β-arrestin1 and p-Src (Figure 5), which based on observations for other GPCRs (13,14), including EP4 (15), could lead to EP2 desensitization and/or provide additional EP2-mediated signaling pathways.

Mutation and activation of H-Ras are considered key events in papilloma development in the initiation/promotion model (16), and the data show that EP2 influenced H-Ras activation in papillomas (Figure 4A and B). These findings suggested that increased levels of PGE₂ in papillomas (18), acting via EP2, could contribute to the activation of H-Ras. Because papillomas contain both WT and mutant H-Ras (28), and mutant H-Ras is considered to exist in the activated form, it is probably that EP2-mediated signaling is primarily influencing the activation of WT H-Ras. However, our data do not allow us to differentiate between EP2’s effects on WT and mutant H-Ras. In support of EP2’s influence on Ras activation, it has been reported that PGE₂ increased Ras activation in rodent and human intestinal cell lines (41,42), and roles for EP1 and/or EP2 in the activation of Ras were indicated (42). Additionally, the data indicated that EP2 contributed to the activation of two H-Ras effectors, ERK1/2 and AKT (Figure 4A and B), both known to be involved in papilloma formation (30,32). However, quantitative differences exist between CAY’s ability to restore the level of these effectors and tumor formation in TPA/Indo-treated mice and the decreased level of these effectors and tumor formed in EP2−/− mice (compare Figure 4A and B). In the former, it appears that EP2 is responsible for total restoration of effector activation and tumor formation, whereas in the latter, EP2 appears to only be responsible for 35–60% of effector activation and ∼65% of the tumors formed. The reasons for CAY stimulation of EP2 showing a more pronounced effect than the extent to which EP2 deficiency decreases these effects are unclear. Strain difference may play a role as the TPA/Indo/CAY effects were observed in promotion-responsive CD-1 mice, whereas the EP2−/− mice were C57BL/6, a strain known to have decreased sensitivity to TPA promotion (43). Notwithstanding, the data are in agreement with EP2 influencing the activation of H-Ras, ERK1/2 and AKT and significantly contributing to tumor formation in the initiation/promotion model.

EGFR is a receptor tyrosine kinase that is known to have a role in skin tumor formation (28,44) and has been demonstrated to activate Ras, ERK1/2, AKT and other signaling effectors (45). The data in Figure 1B demonstrate that AG1478 also reduced papilloma formation in the initiation/promotion model. Of relevance for EP2 contributing to TPA-induced EGFR activation, recent in vitro studies have demonstrated that PGE₂ causes the transactivation of EGFR (22,35,41). In agreement with these effects of PGE₂ on EGFR and reports demonstrating that other GPCRs can cause the transactivation of EGFR (46), the present data showed that CAY stimulation of EP2 contributed to the activation of EGFR (Figure 3D), as well as the activation of H-Ras, ERK1/2 and AKT (Figure 4A). Because EGFR, Ras, ERK1/2 and AKT have been reported to play significant roles in skin tumor development (29,31–33,44), our findings that EP2 contributes to their activation in the initiation/promotion model provides a link between TPA-induced COX-2/PGE₂ (18), EP2-mediated signaling and papilloma development.

Because of the importance of EGFR signaling in papilloma development, studies were initiated to identify possible mechanisms by which EP2 contributed to EGFR activation. Several studies have implicated Src in the transactivation of EGFR by various GPCRs (14,22,35). Based on the observation that CAY stimulation of EP2 contributed to Src activation and that Src was upstream of EGFR (Figure 3D and E), studies were conducted to identify a possible mechanism for EP2’s activation of Src. Several recent reports have indicated that Src can be activated as a result of the formation of a GPCR–β-arrestin1–Src complex (14). The data in Figure 5B show that a complex involving EP2, β-arrestin1 and p-Src could be detected in CAY-treated skin and papillomas by immunoprecipitation of p-Src, suggesting that such a complex could contribute to Src activation in mouse skin. Furthermore, p-Src immunoprecipitation and western blotting of β-arrestin1 and p-Src from WT and EP2−/− mouse skin and papillomas (Figure 5C) indicated that the p-Src–β-arrestin1 complex was decreased in EP2−/− mice, which further indicated EP2’s involvement in the formation of the complex. The immunoprecipitation data in Figure 5C indicate that p-EGFR and p-Src also form a complex, and the β-arrestin1 immunoprecipitation data (Figure 5D) indicate that p-Src disassociates from the EP2–β-arrestin1–p-Src complex prior to interacting with EGFR. However, while our in vivo data indicate the presence of the EP2–β-arrestin1 complex, they do not directly indicate that the complex is involved in Src activation and subsequent transactivation of EGFR or whether the complex is just a mechanism for the downregulation of EP2 by internalization/desensitization. To address this issue, studies with WT and β-arrestin1−/− mice are now in progress. Additionally, it should be stated that while our studies demonstrated the presence of the EP2–β-arrestin1–p-Src complex in mouse skin and papillomas, previous studies utilizing cultured cells suggested that EP2 did not form a complex with β-arrestin following ligand treatment (39,40). Thus, the present data represent the first report of EP2 forming a complex with β-arrestin1 and offer the possibility that the EP2–β-arrestin1 complex may contribute to EP2-mediated signaling.

In addition to EP2 activating the EGFR pathway, we also observed that EP2 activation resulted in increased cAMP levels and the activation of PKA (Figure 3). Regan et al. (10–12) have previously reported that the activation of PKA was a major EP2 signaling pathway and involved activated G_αs stimulation of AC. Our results indicated that EP2-mediated PKA activation led to increased p-CREB levels. Furthermore, PKA has also been shown to phosphorylate glycogen synthase kinase-3β that blocks phosphorylation-dependent β-catenin degradation, thereby allowing β-catenin/TCF-dependent transcription to occur (10). Thus, a major effect of EP2-mediated activation of PKA may be the increased transcription of genes that contribute to transformed keratinocyte growth.

In summary, the results show that the PGE₂ receptor, EP2, activates multiple signaling pathways/effectors that contribute to papilloma formation in the mouse skin initiation/promotion model. Figure 6 illustrates proposed signaling pathways activated by EP2. TPA-induced COX-2 and PGE₂ cause the stimulation of EP2 leading to the G_αs-dependent activation of AC and PKA, as well as G protein-independent activation of Src/EGFR that involves an EP2–β-arrestin1–Src intermediate. While the data illustrate the importance of EP2-mediated signaling in papillomas, it does not identify the cell type in which the signaling was increased nor does it exclude TPA inducing PKA and EGFR activation by pathways independent of EP2 or that EP2 may activate additional pathways. However, the data described herein provide insight into the signaling pathways by which PGE₂ activation of EP2 contributes to skin tumor formation.

Fig. 6.

Proposed model for EP2-mediated signaling. EP2 stimulation by ligand binding leads to G_αs activation and dissociation, which then caused AC activation, cAMP production and PKA activation. EP2 ligand binding also causes EP2 association with β-arrestin1 and Src leading to Src activation. Activated Src then activates EGFR leading the subsequent activation of H-Ras, ERK1/2 and AKT.

Funding

National Institute of Environmental Health Sciences, National Institutes of Health, Intramural program.

Glossary

Abbreviations

AC

adenylate cyclase

AKT

protein kinase B

cAMP

cyclic adenosine 3′,5′-monophosphate

CAY

CAY10399

COX

cyclooxygenase

CREB

cyclic adenosine 3′,5′-monophosphate response element-binding protein

DMBA

7,12-dimethylbenz(a)anthracene

EGFR

epidermal growth factor receptor

ERK

extracellular signal-regulated kinase

GPCR

G protein-coupled receptor

Indo

indomethacin

NSAID

non-steroidal anti-inflammatory drug

PG

prostaglandin

PGE₂

prostaglandin E₂

PKA

protein kinase A

TPA

12-O-tetradecanoylphorbol-13-acetate

WT

wild-type