Prostaglandin E2-EP4 receptor signalling promotes ... - Nature

Oncogene (2007) 26, 3006–3019

& 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc

ORIGINAL ARTICLE

Prostaglandin E2-EP4 receptor signalling promotes tumorigenic behaviour of HT-29 human colorectal cancer cells G Hawcroft, CWS Ko and MA Hull Section of Molecular Gastroenterology, Leeds Institute of Molecular Medicine, University of Leeds, St James’s University Hospital, Leeds LS9 7TF, UK

The predominant product of cyclooxygenase (COX) activity in the colon, prostaglandin (PG) E2 promotes intestinal tumorigenesis. Expression of the PGE2 receptor EP4 is upregulated during colorectal carcinogenesis. Therefore, we investigated the role of elevated PGE2-EP4 receptor signalling in the protumorigenic activity of PGE2 by increasing EP4 receptor expression in HT-29 human colorectal cancer (CRC) cells (HT-29-EP4) by stable transfection. Elevated PGE2-induced EP4 receptor activity in HT-29 cells increased resistance to spontaneous apoptosis and promoted anchorage-independent growth, but had no effect on proliferation of HT-29-EP4 cells. EP4 receptor activation by PGE2 in HT-29-EP4 cells also led to development of fluid-filled cysts, which was associated with increased tight junction protein (occludin and zonula occludens-1) expression. Overexpression of the EP4 receptor in HT-29 cells led to basal EP4 receptor signalling in the absence of exogenous PGE2, which was explained by autocrine activity of endogenous, COX-2-derived PGE2 and constitutive, ligand-independent EP4 receptor activity. The predominant signalling pathway mediating antiapoptotic activity downstream of PGE2-EP4 receptor activation in HT-29-EP4 cells was elevation of cyclic adenosine monophosphate (cAMP) levels, which was associated with phosphorylation of cAMP-response element binding protein. EP4 receptor activation led to a small increase in phosphorylated extracellular signal-regulated kinase (ERK) 2 protein levels but inhibition of ERK phosphorylation did not abrogate the antiapoptotic activity of PGE2. However, PGE2-EP4 receptor signalling did not lead to trans-activation of the epidermal growth factor receptor in HT-29 cells. Inhibition of protumorigenic PGE2-EP4 receptor signalling represents a potential strategy for anti-CRC therapy that may avoid the toxicity associated with systemic COX inhibition. Oncogene (2007) 26, 3006–3019. doi:10.1038/sj.onc.1210113; published online 20 November 2006 Keywords: colorectal cancer; cyclic AMP; EP receptor; prostaglandin E2

Correspondence: Dr G Hawcroft, Section of Molecular Gastroenterology, Leeds Institute of Molecular Medicine, University of Leeds, St James’s University Hospital, Leeds LS9 7TF, UK. E-mail: [email protected] Received 15 May 2006; revised 2 October 2006; accepted 6 October 2006; published online 20 November 2006

Introduction Prostaglandin (PG) E2 is the predominant eicosanoid product of the cyclooxygenase (COX)-PG synthetic pathway in the lower gastrointestinal tract (Krause and DuBois, 2001). PGE2 has been implicated in essential physiological processes in the colon such as electrolyte transport and motility (Krause and DuBois, 2001). PGE2 has also been implicated in the pathogenesis of inflammatory bowel disease, with levels of PGE2 being increased in inflamed tissue compared with uninvolved mucosa (Krause and DuBois, 2001). More recently, a role for PGE2 in intestinal tumorigenesis has been elucidated (Hull et al., 2004). Administration of PGE2 to ApcMin/ þ mice (an established model of the early stages of human colorectal carcinogenesis) has been demonstrated to increase the size and multiplicity of colonic adenomas (Wang et al., 2004). PGE2 is believed to act in an autocrine and/or paracrine manner via a family of four cell surface or nuclear membrane G protein-coupled receptors (GPCRs) termed EP1-EP4, which are linked to distinct second messenger signalling cascades (Breyer et al., 2001). Signalling via both EP2 and EP4 receptors is transduced predominantly by a Gas protein, through which receptor activation is associated with an increase in adenylate cyclase activity and elevated intracellular adenosine-30 , 50 -cyclic monophosphate (cyclic adenosine monophosphate or cAMP) levels (Regan, 2003). However, differences between EP2 and EP4 receptors, including ligand-induced desensitization and internalization (Desai and Ashby, 2001; Fujino et al., 2002), coupling to an inhibitory Gai protein (Fujino and Regan, 2006) and differential signalling via protein kinase (PK) A-independent signal transduction pathways, such as that involving phosphatidylinositol 3kinase (PI3K) and protein kinase B (PKB)/AKT (Regan, 2003), suggest distinct roles for the two ‘stimulatory’ EP receptors. The role of individual EP receptor subtypes at different stages of colorectal carcinogenesis has not been fully characterized. There is genetic and pharmacological evidence implicating a role for Ep1, Ep2 and Ep4 receptors in rodent models of intestinal adenoma and colorectal cancer (CRC) development (reviewed by Hull et al., 2004). Increased levels of Ep1, Ep2 and Ep4 mRNAs, as well as decreased Ep3 transcript levels, are

Protumorigenic activity of PGE2-EP4 receptor signalling G Hawcroft et al

3007

evident in mouse azoxymethane-induced CRC tissue compared with paired normal mucosa (Mutoh et al., 2002; Shoji et al., 2004). More recently, increased levels of EP4 receptor protein have been demonstrated in malignant epithelial cells of human CRCs compared with normal colorectal mucosa (Chell et al., 2006). Previously, investigation of EP receptor activity in CRC cells in vitro has had to rely on the use of poorly selective EP receptor agonists such as PGE1-alcohol and butaprost (Kiriyama et al., 1997; Ma et al., 2006). In this way, the EP4 receptor has been implicated in the PGE2mediated increase in growth, motility and invasive behaviour of human colorectal epithelial cells (Sheng et al., 2001; Chell et al., 2006), as well as PGE2-mediated rescue from the antiproliferative activity of indomethacin on mouse colon 26 CRC cells (Pozzi et al., 2004). As EP4 receptor expression is increased during colorectal carcinogenesis, we investigated the role of PGE2-EP4 receptor signalling cells during colorectal carcinogenesis using a model of EP4 receptor overexpression in HT-29 human CRC cells. Herein, we implicate PGE2 signalling, via elevated EP4 receptor levels, in the protumorigenic progression of intestinal epithelial cells.

Results EP4 receptor mRNA levels, measured by quantitative, real-time polymerase chain reaction (PCR), were increased significantly in human CRC tissue compared with paired normal colorectal mucosa (mean 65% increase (95% confidence interval 29–195%), n ¼ 5; P ¼ 0. 02, one-sample t-test). These data were consistent with the findings from an immunohistochemical study of EP4 receptor protein in human CRCs (Chell et al., 2006) and, together with this previous report, provided the rationale for development of a cellular model of increased EP4 receptor expression in human CRCs. In order to localize cellular EP4 receptor protein accurately, we tagged the human (h) EP4 receptor with a V5 epitope by subcloning hEP4 receptor cDNA into the pTracer vector. Initial experiments testing the function of the hEP4-V5 receptor construct were performed by transient transfection of COS-7 monkey kidney cells and 293T human embryonic kidney cells. Figure 1a demonstrates that hEP4-V5 receptor protein was detectable in total protein lysates of COS-7 and 293T cells transiently transfected with pTracer-hEP4, but not mock transfected cells or COS-7 cells that expressed native, untagged hEP4 receptor from the transfected pCEP4-EP4 vector. Transfection with a pTracer vector expressing LacZ-V5 acted as a positive control (Figure 1a). Both cell types expressed a hEP4-V5 receptor species of approximately 52 kDa but differed in the expression of a larger molecular weight form of EP4 (Figure 1a). The larger form of the EP4 receptor was not detected following lysis of pTracer-hEP4-transfected COS-7 cells in the absence of detergent, but was still present in a pTracer-hEP4-transfected 293T-cell lysate produced in a

similar manner (data not shown), suggesting that, in COS-7 cells, this species represents a membrane-bound form of the receptor that requires detergent-solubilization for detection by Western blotting. Immunofluorescence on hEP4-V5 receptor-transfected COS-7 cells in the presence of solvent carrier alone demonstrated prominent hEP4-V5 immunoreactivity at the cell membrane (Figure 1b). However, upon stimulation with 1 mM PGE2 for 60 min, membranous hEP4-V5 receptor immunoreactivity became almost exclusively cytoplasmic (Figure 1b). As expected, PGE2 signalling via the V5-tagged hEP4 receptor in 293T cells, in the presence of the phosphodiesterase IV inhibitor rolipram, was associated with a greater than 10-fold increase in intracellular cAMP levels (Figure 1c). The PGE2induced rise in cAMP levels mediated by the hEP4-V5 receptor was of similar magnitude to that seen in cells transfected with native EP4 (pCEP4-hEP4; Figure 1c). Transfection of 293T cells with pTracer-hEP4 was also associated with a PGE2-mediated increase in b-catenin/ T-cell factor (TCF)-mediated transcription (Figure 1d), consistent with previously published data (Fujino et al., 2002). Taken together, these data confirm intact EP4 receptor trafficking and downstream signalling following transfection with pTracer-hEP4. Having confirmed the functionality of the hEP4-V5 receptor, we chose HT-29 human CRC cells to produce EP4 receptor overexpressing colorectal epithelial cells, thereby mirroring the increase in epithelial cell EP4 protein levels that occurs in human CRCs (Chell et al., 2006). The HT-29 cell line was particularly suitable for experiments testing PGE2-EP4 receptor signalling and function as HT-29 cells do not produce detectable endogenous PGE2 (Hsi et al., 2000; Shao et al., 2000), do not demonstrate an elevation in intracellular cAMP levels upon addition of exogenous PGE2 (Cassano et al., 2000) and contain EP4 receptor mRNA, but not transcripts for the other EP receptor subtypes coupled to cAMP signalling (Cahlin et al., 2005; Colucci et al., 2005). HT-29 cells were transfected with the pTracerhEP4 vector in the presence of the selection agent blasticidin. Stably transfected EP4 receptor overexpressing (HT-29-EP4) and empty vector-transfected (HT-29con) single-cell clones were then obtained by limited dilution cloning and culture in the continued presence of blasticidin. Untreated EP4 receptor-overexpressing HT-29-EP4 cells exhibited the same membranous distribution of V5tagged EP4 receptor as transiently transfected COS-7 cells (Figure 2a and b). HT-29-EP4 cells also underwent ligand (PGE2)-induced EP4 receptor internalization (Figure 2c). By contrast, there was no specific V5 immunofluorescence detected in HT-29-con cells (Figure 2d), thus confirming the specificity of the V5 immunoreactivity in HT-29-EP4 cells. In keeping with data from Cassano et al. (2000), cAMP was not detected in HT-29-con cells even after exposure to 1 mM PGE2 for 5 min (data not shown). However, HT-29-EP4 cells exhibited a marked elevation of intracellular cAMP levels upon stimulation with PGE2, at concentrations as low as 50 nM, for 5 min Oncogene


3008

b

a COS-7

1

2

3

4 5

293T

6

7

PGE2

carrier

8 97 64

IB:V5 51

IB:β-actin

c

42

3500 Rolipram alone 3000

Rolipram + carrier

d

Rolipram + PGE2

pTOPflash luciferase activity

cAM P (fmol/well)

2500 2000 1500 1000 500 0

1.6 1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 -

pTracer

pTracer-hEP4

+

-

+

pCEP4-EP4 pTracer

pTracer-hEP4

Figure 1 Expression and functionality of V5-tagged hEP4 protein in COS-7 and 293T cells. (a) Immunoblot analysis of the V5 epitope and b-actin protein. Total protein lysates were obtained from COS-7 cells (lanes 1–6) and 293T cells (lanes 7–8) in the presence (lanes 1–5 and 7–8) or absence (lane 6) of 1% (v/v) Brij-96 (detergent). Lane 1, mock transfection; lane 2, pTracer-hEP4 (detergent); lane 3, pCEP4-EP4; lane 4, pTracer-lacZ; lane 5, empty pTracer vector; lane 6, pTracer-hEP4 (no detergent); lane 7, mock transfection; lane 8, pTracer-hEP4 (detergent). Figures on the right denote the size of bands in kDa. (b) Indirect imunofluorescence for V5 on pTracerhEP4-transfected COS-7 cells treated with either 0.1% (v/v) DMSO carrier alone (carrier) or 1 mM PGE2 for 1 h. V5 immunoreactivity was localized predominantly at the surface membrane of carrier-treated cells (solid arrow) but underwent ligand-induced translocation to the cytoplasm in PGE2-treated cells (dashed arrow). The inset demonstrates the lack of V5 immunoreactivity in empty pTracer vector-transfected cells. No specific fluorescence was observed when incubation with the primary anti-V5 antibody was omitted (data not shown). Nuclear localization with DAPI. Scale bar ¼ 10 mm. (c) cAMP levels in 293T cells transiently transfected with either empty pTracer vector, pTracer-hEP4 or pCEP4-EP4. Cells were pretreated with 50 mM rolipram for 10 min before addition of 1 mM PGE2 or an equivalent dilution (0.1% v/v) of ethanol carrier alone for 5 min before cell harvesting. Data represent the mean and s.e.m. of five replicates. (d) b-catenin/TCF-related transcription measured by dual luciferase assay of pTOPflash activity in 293T cells transiently transfected with either empty pTracer vector or pTracer-hEP4. Cells were treated with 1 mM PGE2 ( þ ) or an equivalent dilution of carrier alone () for 1 h and then incubated for 16 h before lysis. Data represent the mean and s.e.m. of triplicate experiments measuring pTOPflash luciferase activity normalized to Renilla luciferase values, in order to correct for transfection efficiency.

(Figure 2e). Exposure of HT-29-EP4 cells to 1 mM PGE2, a concentration at which PGE2 has been used widely in previous in vitro studies, was not associated with a further stepwise increase in the cAMP response of HT-29-EP4 cells above that seen with 50 nM PGE2 (Figure 2e). Further evidence implicating the EP4 receptor in the differential cAMP response of stably transfected HT-29-EP4 cells was obtained by demonstrating that the EP2 receptor-selective agonist butaprost did not induce an increase in cAMP levels in HT-29-EP4 cells, unlike PGE1-alcohol (PGE1-OH), which at low concentrations (10 nM), is a relatively Oncogene

specific agonist for the EP4 receptor (Kiriyama et al., 1997 Figure 2e). We also tested the EP4 receptor antagonist ONO-AE3-208 in this assay system and determined that approximately 90% inhibition of the cAMP response of HT-29-EP4 cells to 1 mM PGE2 occurred with 10 mM ONO-AE3-208 (Figure 2f), a concentration consistent with that used in a previous in vitro study using mouse macrophages (Pavlovic et al., 2006). There was a small elevation in cAMP levels observed in basal (non-PGE2treated), rolipram-treated HT-29-EP4 cells compared with HT-29-con cells (Figure 2e) and ONO-AE3-208 also decreased intracellular cAMP levels in HT-29-EP4


3009

a

b

c

d

e 22000

f 100

20000 18000

80 % decrease in cAMP

cAMP (fmol/well)

16000 14000 12000 10000 8000 6000

60

40

20

4000 2000 0

0 carrier

E2 50

E21

but

E1-OH

-

+

Figure 2 Expression of the EP4 receptor in HT-29 cells. (a) Indirect immunofluorescence for V5 in untreated HT-29-EP4 cells showing predominant membranous localization of the V5-tagged hEP4 receptor. Nuclear localization with DAPI. No specific fluorescence was observed when incubation with the primary anti-V5 antibody was omitted (data not shown). Size bar ¼ 20 mm. (b) Higher magnification view of membranous hEP4-V5 immunoreactivity in untreated HT-29-EP4 cells. Nuclear localization with DAPI. Size bar ¼ 20 mm. (c) Indirect immunofluorescence for V5 in HT-29-EP4 cells treated with 1 mM PGE2 for 1 h showing ligandinduced internalization of the V5-tagged hEP4 receptor. Nuclear localization with DAPI. Scale bar ¼ 20 mm. (d) Indirect immunofluorescence for V5 on untreated HT-29-con cells. There was no specific signal for V5 confirming the specificity of the V5 localization in HT-29-EP4 cells (compare with a–c). Nuclear localization with DAPI. Scale bar ¼ 20 mm. (e) Intracellular cAMP levels in HT-29-EP4 cells treated with EP receptor agonists PGE2 (E2; 50 nM and 1 mM), butaprost (but; 1 mM) and PGE1-alcohol (E1-OH;10 nM) for 5 min, following pretreatment with 50 mM rolipram. Data represent the mean and s.e.m.; n ¼ 3 for each condition. (f) Percentage decrease in intracellular cAMP levels in rolipram-treated HT-29-EP4 cells following preincubation with ONO-AE3-208 for 45 min before treatment with carrier alone () or 1 mM PGE2 ( þ ) for 5 min. Data represent the mean and s.e.m.; n ¼ 3 for each condition.

Oncogene


3010

cells in the absence of exogenous PGE2 (Figure 2f). One possible explanation is that HT-29-EP4 cells produce PGE2, which could drive EP4 receptor activity in the absence of exogenously added PGE2. However, PGE2 was not detectable by conventional immunoassay of medium conditioned by HT-29-EP4 (or HT-29-con) cells, even after treatment with the EP4 receptor agonist PGE1-OH (10 nM or 1 mM). Alternatively, we used the non-selective COX inhibitor indomethacin and the selective COX-2 inhibitor SC236, at non-toxic concentrations (Smith et al., 2000; Ko et al., 2002), in order to determine whether low level COX-dependent PGE2 synthesis, below the detection limit of the immunoassay, accounted for basal EP4 receptor activity in HT-29-EP4 cells. Incubation with indomethacin and SC236 (both 1 mM) for 30 min was associated with a 54.5712.7 (standard error of the mean (s.e.m.)) % and 60.0710.0% decrease in intracellular cAMP levels, respectively, in HT-29-EP4 cells (n ¼ 3 in two independent experiments). Importantly, prior incubation with a 10-fold higher concentration of indomethacin did not produce any further inhibition of cAMP production in HT-29-EP4 cells (47.877.7%) suggesting that the response to 1 mM concentrations of the COX inhibitors was maximal. These data suggest that ligand-dependent EP4 receptor activation by low levels of PGE2 (below the detection level of a high-sensitivity immunoassay) is responsible for a proportion (approximately 60%) of basal EP4 receptor activity in HT-29-EP4 cells and that COX-2 is the predominant COX isoform responsible for trace PGE2 synthesis in HT-29-EP4 cells. Overall, the immunolocalization and cAMP data confirmed that the transfected hEP4 receptor was also functional in HT-29 cells and provided the impetus to study further the phenotype of HT-29-EP4 cells. The most obvious phenotypic change in untreated HT-29-EP4 cells was the presence of cysts in the HT-29 multi-cell layer (Figure 3a), which were not evident in untreated HT-29-con (Figure 3b) or parental HT-29 cell cultures (data not shown). Formation of cysts was a dynamic process with the expansion and disappearance of individual cysts over time demonstrated by time-lapse photomicroscopy (see Supplementary video file Figure S1). Mucin histochemistry revealed that the cysts were Alcian Blue (pH 2.5)-negative and Periodic Acid-Schiffnegative (data not shown). We hypothesized that the cysts represented accumulation of fluid within the cell multilayer due to increased electrolyte transport and/or decreased paracellular

permeability of HT-29-EP4 cell cultures. Confocal immunofluorescence for tight junction (TJ) proteins zonula occludens (ZO)-1 and occludin demonstrated coordinated upregulation of both TJ proteins in cells surrounding the cyst spaces (Figure 3c). Expression of ZO-1 and occludin was largely restricted to cells surrounding the cysts with only small areas of discontinuous ZO-1 and occludin immunoreactivity being apparent in non-cystic areas of HT-29-EP4 and HT-29con cell layers (see Supplementary Figure S2). Electron microscopy demonstrated that TJs and desmosomes were prominent in HT-29-EP4 cells (Figure 3d). In order to determine further the mechanistic basis of cyst formation we developed a quantitative, photomicroscopic assay of the cyst cross-sectional area measured at any given time. As described previously (Ophir et al., 1995), the adenylate cyclase activator forskolin induced cyst formation in HT-29-con cells (Figure 3e) implying a role for cAMP signalling. However, PGE2 did not promote visible cyst formation in HT-29-con cells (data not shown). By contrast, PGE2 increased cyst formation in HT-29-EP4 cells (to a similar degree to forskolin), which was inhibited by ONO-AE3-208, thereby implicating PGE2-EP4 receptor-cAMP signalling in cyst formation in HT-29 cell cultures in vitro. Importantly, the Na þ /K þ -ATPase inhibitor ouabain inhibited basal and PGE2-induced cyst formation almost completely in HT-29-EP4 cells (Figure 3e) implying that active electrolyte transport is necessary for EP4 receptordependent cyst formation in HT-29-EP4 cells, in combination with increased TJ protein expression. As reduced apoptosis is a hallmark of carcinogenesis (Wright and Duckett, 2005) and cAMP signalling is recognized to promote resistance to apoptosis in vitro (Hoshino et al., 2003; Nishihara et al., 2004), we next investigated the role of PGE2-EP4 receptor signalling in the apoptotic potential of HT-29 cells. Adherent human CRC cells, including HT-29 cells, become detached during apoptosis (Elder et al., 1997; Gardner et al., 2004). Therefore, we used the established technique of non-adherent cell counting in order to quantitate the degree of apoptosis (Elder et al., 1997; Gardner et al., 2004). Preliminary experiments confirmed that the nonadherent cell count accurately reflected the degree of apoptosis in stably transfected HT-29 cell cultures (mean (s.e.m.) 98.4 (0.4) % non-adherent HT-29-con cells and 98.8 (0.7) % non-adherent HT-29-EP4 cells (both n ¼ 6) had apoptotic morphology by fluorescence microscopy). Exogenous PGE2 (1 mM) had no effect on

Figure 3 Intercellular cyst formation by HT-29-EP4 cells. (a) Phase-contrast photomicrograph of untreated HT-29-EP4 cells showing multiple translucent cysts (arrowheads). Size bar ¼ 250 mm. (b) Phase-contrast photomicrograph of untreated HT-29-con cells showing typical HT-29 cell morphology, with no evidence of cyst formation. Size bar ¼ 250 mm. (c) Indirect immunofluorescence for TJ proteins ZO-1 (red) and occludin (green) on 1 mM PGE2-treated HT-29-EP4 cells (24 h) demonstrating membranous colocalization (yellow) in cells surrounding cyst structures (asterisks). Size bar ¼ 100 mm. (d) Transmission electron photomicrograph of untreated HT-29-EP4 cells showing a cyst in cross-section (asterisk) surrounded by cells with prominent TJs and desmosomes (arrowheads), which were not evident in HT-29-con cells. The high power inset highlights a prominent brush border in HT-29-EP4 cells. (e) Quantitative analysis of the total cyst area of HT-29 cells measured by phase-contrast microscopy. HT-29-con or HT-29-EP4 cells were exposed to DMSO carrier (0.1% v/v) alone, PGE2 (10 mM), forskolin (Fsk; 15 mM) or no treatment (culture medium alone) for 24 h. Some cells underwent preincubation for 60 min with 10 mM ONO-AE3-208 (EP4A) or ouabain (oua; 10 mM) before 24 h incubation. The total cyst area was measured in four separate, quadrantic microscopic fields of a single well for each experimental condition. Oncogene


3011

HT-29-con cell apoptosis (Figure 4a). Untreated HT-29EP4 cells displayed reduced basal apoptosis compared with HT-29-con cells (Figure 4a). By contrast with HT29-con cells, 1 mM PGE2 reduced spontaneous apoptosis of HT-29-EP4 cells by approximately 60% (Figure 4a).

Similar findings were obtained when we measured apoptosis in HT-29 cells by flow cytometry for Annexin V (A) and propidium iodide (PI) staining. The mean percentage number of A þ , PI (indicative of early stage apoptosis) HT-29-EP4 cells cultured in the

a

b

c

d

*

*

e

*

14000 13000 12000

total cyst area (µm2)

11000 10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0

HT-29-con

+

+

-

-

-

-

-

-

-

-

HT-29-EP4

-

-

+

+

+

+

+

+

+

+

PGE2

-

-

-

-

-

-

-

+

+

+

carrier alone

+

-

-

+

-

-

-

-

-

-

EP4A

-

-

-

-

+

-

-

-

+

Fsk

-

+

-

-

-

-

+

-

-

-

Oua

-

-

-

-

-

+

-

-

-

+ Oncogene


3012

a

14 13

No. of apoptotic cells (x 104)

12 11 10 9 8 7 6 5 4 3 2 1 0

b

55

HT-29-con

+

+

-

-

-

-

HT-29-EP4

-

-

+

+

+

+

carrier alone

+

-

+

-

-

-

PGE2

-

+

-

+

-

+

EP4A

-

-

-

-

+

+

55

HT-29-con

50

45 cell number (x 104)

cell number (x 104)

HT-29-EP4

50

45 40 35 30 25 20 15

40 35 30 25 20 15

10

10

5

5

0

0 24

48 time (h)

72

c

24

HT-29-con

48 time (h)

72

HT-29-EP4

relative DNA synthesis (OD450)

0.25

0.20

0.15

0.10

0.05

0.00 24

48 time (h)

72

24

48

72

time (h)

Figure 4 Antiapoptotic activity of EP4 receptor signalling in HT-29 cells. (a) Spontaneous apoptosis in post-confluent HT-29-con and HT-29-EP4 cells following treatment with DMSO carrier or 1 mM PGE2 for 24 h with or without preincubation for 60 min with 10 mM ONO-AE3-208 (EP4A). Data represent the mean and s.e.m.; n ¼ 3. Similar results were obtained in four independent experiments. (b) Counting of adherent, viable HT-29-con and HT-29-EP4 cells over 72 h in the presence of DMSO (0.1% v/v) alone (solid squares), 1 mM PGE2 (open squares) or 10 mM ONO-AE3-208 (solid triangles). Data represent the mean and s.e.m. of triplicate values for each time point. (c) Proliferation of HT-29-con and HT-29-EP4 cells measured by BrdU incorporation assay in the presence of DMSO (0.1% v/v) alone (open bars) or 1 mM PGE2 (shaded bars). Data represent the mean and s.e.m. of triplicate optical density (OD) values at 450 nm for each time point between 24 and 72 h.

presence of dimethylsulphoxide (DMSO) carrier alone, in two independent experiments, was 1.25% compared with 2.25% of HT-29-con cells. Moreover, PGE2 (1 mM) reduced the percentage number of spontaneously apoptotic A þ , PI HT-29-EP4 cells by 24 to 0.95% of the total cell number. In HT-29-EP4 cells, the antiapoptotic effect of PGE2 was antagonized by ONO-AE3-208 (Figure 4a). In Oncogene

keeping with data from the cAMP assay, inhibition of the antiapoptotic effects of PGE2 on HT-29-EP4 cells by ONO-AE3-208 only occurred at a concentration of 10 mM, unlike lower concentrations (10 nm–1 mM), which did not reverse PGE2 activity in our non-adherent cell counting assay (data not shown). Taken together, these data implicate elevated EP4 receptor levels in PGE2mediated protection of HT-29 cells from apoptosis.


3013

a

b

c

3.5

% colony forming efficiency

3.0 2.5 2.0 1.5

carrier

carrier

PGE2

PGE2

PGE2 + EP4A

PGE2 + EP4A

1.0 0.5 0.0

HT-29-con + + + HT-29-EP4 - carrier alone - + EP4A PGE2

-

-

- -

-

-

-

-

+ + + + + - + - - -

+

- - + - + - - - + +

Figure 5 Promotion of anchorage-independent growth of HT-29 cells by PGE2-EP4 receptor signalling. (a) The number of colonies of HT-29-con and HT29-EP4 cells in soft agar at day 14 measured as the % colony forming efficiency. (b) Phase-contrast microscopy of HT-29-EP4 cells following treatment with carrier (DMSO) alone, 1 mM PGE2 or 1 mM PGE2 þ 10 mM ONO-AE3-208 (EP4A) for 14 days. Note the intercellular cysts present in HT-29-EP4 cell colonies cultured in the presence of 1 mM PGE2. (c) Macroscopic views of HT-29-EP4 cell colonies following treatment with carrier alone, 1 mM PGE2 or 1 mM PGE2 þ 10 mM ONO-AE3-208 (EP4A) for 14 days.

By contrast, exogenous PGE2 (1 mM) did not increase the proliferation rate of either HT-29-con or HT-29-EP4 cells over a period of 72 h as measured by adherent cell counting (Figure 4b) or a DNA synthesis assay (Figure 4c). The number of adherent, viable, untreated HT-29-EP4 cells increased more quickly over time than PGE2-untreated HT-29-con cells (Figure 4b). However, proliferation, as measured by 5-bromo-20 -deoxyuridine (BrdU) incorporation, did not differ between PGE2untreated HT-29-EP4 cells and HT-29-con cells (Figure 4c). Overall, in this human CRC model, it can be concluded that increased PGE2-EP4 receptor activity does not stimulate a significant proliferative response in HT-29 cells. Another cardinal feature of tumorigenic behaviour of transformed cells is the ability to grow in an anchorageindependent manner. Control-transfected HT-29 cells exhibited low efficiency anchorage-independent growth in soft agar (Figure 5), which was not affected by treatment with PGE2 (Figure 5). However, untreated HT-29-EP4 cells demonstrated a greater than fourfold increase in cell colony formation, which was increased further in the presence of exogenous PGE2 (Figure 5). Importantly, ONO-AE3-208 reduced the colony-forming efficiency and colony size of untreated and PGE2-treated HT-29-EP4 cells (Figure 5). These data suggest a role for increased EP4 receptor signalling in promotion of anchorage-independent growth of HT-29 cells. Several signalling pathways have been described downstream of EP4 receptor activation, including

protein kinase A (PKA) activation, PI3K signalling and extracellular signal-regulated kinase (ERK) activation, in transfected EP4 receptor-expressing 293T HEK cells (Fujino et al., 2002, 2003, 2005). However, the relevance of these findings to malignant human colorectal epithelial cells with increased EP4 receptor expression is unclear. Therefore, we studied signalling downstream of EP4 receptor activation in HT-29 human CRC cells. As elevation of cAMP levels was a prominent feature of PGE2-induced activation of the EP4 receptor in HT-29-EP4 cells, we firstly examined activation of the transcription factor cAMP-response element binding protein (CREB) in response to EP4 receptor activation by PGE2. HT-29-con cells contained only low levels of phosphorylated (p)CREB protein and the phosphorylated form of the related CREB family member activating transcription factor 1 (ATF1) relative to the total intracellular CREB level (Figure 6a). In a pattern consistent with the cAMP response to PGE2 treatment of HT-29-con and HT-29-EP4 cells (Figure 2e), PGE2 did not induce phosphorylation of CREB/ATF1 in HT-29-con cells but increased pCREB/pATF1 levels in HT-29-EP4 cells in an EP4 receptor-dependent manner (Figure 6a). Phosphorylation of CREB following EP4 receptor activation by PGE2 implicates PKA signalling downstream of the EP4 receptor (Mayr and Montminy, 2001). However, the PKA inhibitor H89 only partially reversed (by approximately 50%) PGE2-induced resistance to apoptosis of HT-29-EP4 cells, suggesting a Oncogene


3014

b 2

3

ve

HT-29-EP4

HT-29-con 1

4

1

2


a iti

3

4

pCREB pATF1

s Po

IB:pCREB IB:total CREB

5 4 3

*

2 1 0

carrier alone PGE2 H89

1 ERK1 ERK2 ERK1

d

e

tiv

2

3

4

5

6

si Po

IB:pERK1/2 IB:total ERK1/2

ERK2


c

+ -

+

+

-

-

+

5 4 3 2 1 0

carrier alone + PGE2 -

+

+

+

PD98059

-

-

+

-

U0126

-

-

-

+

Figure 6 EP4 receptor-mediated signalling in HT-29 cells. (a) Immunoblot analysis of pCREB/pATF-1 and total CREB protein levels in HT-29-con and HT-29-EP4 cells. Cells were serum-starved for 24 h before addition of 10 mM ONO-AE3-208 (EP4A) for 60 min before addition of PGE2 (1 mM) or DMSO carrier (0.05% v/v dilution), 60 min before cell lysis. Lane 1, carrier alone; lane 2, PGE2; lane 3, EP4A þ PGE2; lane 4, EP4A alone. Positive denotes a lysate of SK-N-MC cells treated with forskolin and 3-isobutyl-1methylxanthine. (b) The effect of the PKA inhibitor H89 on HT-29-EP4 cell apoptosis. H89 (10 mM) or DMSO carrier alone were preincubated with cells for 60 min before addition of PGE2 (1 mM) for 24 h. Data represent the mean and s.e.m.; n ¼ 3. *P ¼ 0.04 for the comparison between PGE2-treated and PGE2 þ H89-treated cells (Student’s unpaired t-test). Data are representative of two independent experiments. (c) Immunoblot analysis of pERK1/2 and total ERK1/2 protein levels in HT-29-EP4 cells. Cells were serumstarved for 24 h before addition of ONO-AE3-208 or H89 (10 mM) for 60 min before addition of PGE2 (1 mM) or DMSO carrier (0.05% v/v dilution), 60 min before cell lysis. Lane 1, carrier alone; lane 2, PGE2; lane 3, EP4A þ PGE2; lane 4, EP4A alone; lane 5, H89 þ PGE2; lane 6, H89 alone; positive, 100 ng/ml rhEGF for 5 min. Data are representative of two independent experiments. (d) The effect of MEK inhibition on HT-29-EP4 cell apoptosis. PD98059 (50 mM) or U0126 (25 mM) were preincubated with cells for 60 min before addition of PGE2 (1 mM) for 24 h. Data represent the mean and s.e.m.; n ¼ 3. There was no significant difference (P>0.05; Student’s unpaired t-test) between the degree of apoptosis of PGE2-treated HT-29-EP4 cells in the absence or presence of either PD98059 or U0126. The MEK inhibitors did not alter apoptosis of HT-29-EP4 cells that were not treated with PGE2 (data not shown).

contribution from a PKA-independent signalling pathway towards EP4 receptor-mediated antiapoptotic activity in HT-29 cells (Figure 6b). One candidate PKA-independent pathway is ERK activation, as previous studies have described EP4 receptor-dependent ERK signalling in mouse colon 26 CRC cells (Pozzi et al., 2004) and PGE2-induced ERK signalling in HCA-7 human CRC cells (Wang et al., 2005). Treatment with PGE2 did not alter the cellular levels of phosphorylated or total ERK1 and ERK2 in HT-29-con cells (data not shown). However, treatment with 1 mM PGE2 increased pERK2 levels, relative to the total ERK1/2 content, in HT-29-EP4 cells at 60 min (Figure 6c). This effect was abrogated in the presence of ONO-AE3-208 (Figure 6c). EP4 receptor-mediated induction of pERK2 was also inhibited by 10 mM H89 (Figure 6c), implying that ERK activation is, in fact, linked to PKA activation in HT-29 cells. By contrast, incubation with 100 ng/ml rhEGF for 5 min was associated with a far greater induction of pERK in HT-29 cells (Figure 6c). In keeping with relatively minor pERK induction downstream of EP4 receptor activation Oncogene

in HT-29 cells overexpressing the EP4 receptor, preincubation with either of two MAP kinase (MEK) inhibitors PD98059 and U0126 did not abrogate the antiapoptotic activity of PGE2-EP4 receptor signalling in HT-29-EP4 cells (Figure 6d). PGE2 has been implicated in trans-activation of the epidermal growth factor receptor (EGFR) in several cell types in vitro, including human CRC cells (Pai et al., 2002; Buchanan et al., 2003, 2006). Therefore, we used our HT-29 human CRC model in order to investigate the involvement of the EP4 receptor in PGE2-mediated EGFR trans-activation. Contrary to a previous report (Pai et al., 2002), we did not observe trans-activation of EGFR by PGE2 incubated for varying times (5–60 min) with HT-29-con cells (Figure 7a and b) using two independent techniques (phospho-tyrosine (PY) detection in immunoprecipitated EGFR and cell-based immunoassay of pEGFR), although this cell line does express functional EGFR, as its direct ligand EGF induced receptor phosphorylation in both assays (Figure 7a and b). Furthermore, in order to confirm that the lack of PGE2-mediated EGFR trans-activation was


3015

a

HT-29-EP4

HT-29-con 1

2

3

4

1

2

3 IB:PY

IP:EGFR IB:EGFR

1.6

1.5

1.5

1.4

1.4

1.3

1.3 1.2

µM 10

1

2

PG

E

2

E

er

at tre no

µM

on

m

en

µM 2

E

2

PG

PG

E

1

al er rri ca

10

µM

on

F EG

en m at tre no

HT-29-con

PG

0.8 al

0.8

t

0.9

e

1.0

0.9

e

1.1

1.0

rri

1.1

ca

1.2

pEGFR total EGFR

EG F

OD450

1.6

t

OD450

b

HT-29-EP4

Figure 7 PGE2-EP4 receptor signalling does not trans-activate the EGFR in HT-29 cells. (a) Immunoprecipitation of EGFR from HT-29-con and HT-29-EP4 cell lysates followed by immunoblot analysis of PY and EGFR. Cells were serum-starved for 24 h beforehand. Lane 1, no treatment; 2, DMSO carrier alone (0.1% v/v); 3, 1 mM PGE2 for 15 min; 4, 100 ng/ml rhEGF for 5 min. Similar results were obtained following incubation with 1 mM PGE2 for 5 min or 60 min. (b) Cell-based immunoassay of pEGFR (tyr1173) and total EGFR levels in HT-29-con and HT-29-EP4 cells. Cells were serum-starved for 24 h before addition of rhEGF (100 ng/ml) for 5 min, DMSO carrier alone (0.1% v/v) for 15 min or PGE2 (1 mM for 15 min or 10 mM for 60 min). OD values at 450 nm were normalized to cell number by crystal violet absorbance at 595 nm. Data represent the mean and s.e.m.; n ¼ 2. Solid bars represent pEGFR levels and open bars denote total EGFR levels. Results were similar in two independent experiments.

not due to selection of an EGFR-null cell clone, we demonstrated that PGE2 did not induce EGFR phosphorylation in non-transfected parental HT-29 cells (data not shown). Similarly, treatment with 1 mM PGE2 for 15 min, or a 10-fold higher concentration of PGE2 for 60 min, did not increase EGFR phosphorylation in HT-29-EP4 cells (Figure 7a and b). Therefore, we have found no evidence that the EP4 receptor mediates PGE2-mediated EGFR trans-activation in HT-29 cells. Consistent with the absence of PGE2-induced EGFR activation, we did not observe an increase in pPKB/ AKT levels in HT-29-EP4 cells treated with PGE2 (data not shown). This suggests that this important cell survival signalling pathway does not mediate the protumorigenic activity of the EP4 receptor in HT-29 cells. Discussion Recent clinical data suggest that long-term selective COX-2 inhibition is associated with an increased risk of myocardial infarction and stroke (Grosser et al., 2006). However, COX-2 inhibition has potent antineoplastic activity (Hull, 2005). Therefore, it is important to develop strategies that harness the beneficial anticancer effects of COX inhibition, while avoiding the vascular toxicity associated with systemic COX-2 inhibition. Our findings suggest that inhibition of PGE2-EP4 receptor signalling may be a valid anti-CRC strategy, particularly

as preclinical studies have suggested that EP4 receptor antagonism may have a favourable side effect profile (Kato et al., 2005; Pavlovic et al., 2006). Some phenotypic changes, including cyst formation, increased anchorage-independent growth and elevated intracellular cAMP levels, were evident in HT-29-EP4 cells in the absence of exogenously added PGE2. Experiments with COX inhibitors demonstrated that approximately 60% of the basal EP4 receptor activity in PGE2-untreated HT-29-EP4 cells, as measured by the cAMP response, was explained by autocrine effects of COX-2-derived PGE2. One explanation for the increased cAMP signalling in HT-29-EP4 cells, which was not abrogated by COX-2 inhibition, is that increased expression of the EP4 receptor in HT-29EP4 cells may have uncovered a degree of constitutive, ligand-independent EP4 receptor activity, similar to that described for many other GPCRs (Seifert and WenzelSeifert, 2002). Potential PGE2-independent EP4 receptor activity occurring in colorectal tumours exhibiting elevated levels of the EP4 receptor has important implications for the efficacy of anticancer therapies targeting PGE2 synthesis, for example, COX inhibitors. In our model of EP4 receptor overexpression in HT-29 cells, PGE2-EP4 receptor signalling was linked predominantly to cAMP signalling and low level ERK activation, but not PKB/AKT signalling, in contrast to findings from a model of EP4 receptor expression in 293T cells, in which cAMP signalling appears to play a relatively minor role in signal transduction (Fujino Oncogene


3016

et al., 2002, 2003, 2005). Therefore, this study highlights the importance of investigating EP receptor signalling in a relevant cellular context. EP4 receptor-dependent antiapoptotic activity of PGE2 in HT-29 cells occurred via a mechanism involving cAMP-PKA signalling but not ERK activation or PI3K-PKB/AKT signalling. By contrast, cAMP-dependent suppression of apoptosis by PGE2 in T84 human CRC cells seems to occur via mechanisms dependent on ERK and p38 mitogenactivated (MA) PK signalling, but not PKA (Nishihara et al., 2004). There is clearly significant heterogeneity in signalling pathways mediating the antiapoptotic and proproliferative activities of PGE2 between different CRC cells in vitro (Sheng et al., 2001; Nishihara et al., 2004; Pozzi et al., 2004; Holla et al., 2005). This may be due to interplay between the EP receptor subtypes that are variably present on different CRC cell lines. For example, HT-29 and T84 human CRC cell lines both express EP4 receptor transcripts (Colucci et al., 2005; Cahlin et al., 2005, Nishihara et al., 2003), but T84 cells also contain EP2 receptor mRNA, unlike HT-29 cells (Nishihara et al., 2003). PKA activation only accounted for approximately 50% of the antiapoptotic activity related to PGE2-EP4 receptor signalling in HT-29 cells. It is increasingly recognized that cAMP can also signal in a PKAindependent manner via the cAMP-dependent guanine nucleotide exchange factor Epac1, which, in turn activates the Ras-GTPase Rap1 (Misra and Pizzo, 2005). It will be important to define the role of Epac1Rap1 signalling downstream of EP4 receptor activation in human CRC cells. PGE2-dependent EGFR activation in human CRC cells also appears to be variable, with responsive (LS174) and unresponsive (DLD-1) cell lines described (Buchanan et al., 2003, 2006; Castellone et al., 2005). In our study, we found no evidence of PGE2-induced transactivation of the EGFR in HT-29 cells, which others have noted express constitutively active EGFR (Keese et al., 2005). PGE2-mediated EGFR trans-activation in other human cancer cell lines involves other EP receptor subtypes. For example, the EP1 receptor has been linked to EGFR phosphorylation in CCLP1 human cholangiocarcinoma cells (Han and Wu, 2005) and Hep3B human hepatocellular carcinoma cells (Han et al., 2006). Moreover, PGE2 has been demonstrated to transactivate the EGFR via the EP2 receptor in Ishikawa human endometrial cancer cells (Sales et al., 2004). However, the EP4 receptor may play a role in EGFR trans-activation in some cell types as the EP4 receptor antagonist L161982 abrogated PGE2-induced EGFR phosphorylation in rat cardiac myocytes (Mendez and LaPointe, 2005). Overexpression of the EP4 receptor in HT-29 cells has also provided novel insights into the relationship between PGE2 signalling and paracellular permeability and the ion transport properties of intestinal epithelial cell cultures. Our data implicate EP4 receptor activation in TJ assembly in HT-29 cells that normally do not exhibit TJs (Ophir et al., 1995). Interestingly, cAMP has been demonstrated to increase ZO-1 and occludin Oncogene

protein levels in human microvascular endothelial cells (Dye et al., 2001) and TJ multiplicity in intact mesenteric blood vessels in frogs (Adamson et al., 1998), although there has been no previous study linking cAMP signalling and TJ formation in epithelial cells. In summary, we have demonstrated that increased PGE2EP4 receptor signalling that occurs in CRCs drives protumorigenic behaviour of HT-29 cells, via a mechanism that involves cAMP-PKA signalling, but not ERK activation or AKT phosphorylation downstream of EGFR activation. By comparison with disparate sources of data from studies using other cell lines (Sheng et al., 2001; Buchanan et al., 2003; Hoshino et al., 2003; Nishihara et al., 2004), it is clear that similar phenotypic responses to exogenous PGE2 can be subserved by different signalling pathways acting in a cell-specific manner. In future, it will be important to determine the complete EP receptor profile, the PGE2 content and the degree of ligand-independent EP receptor activity of colorectal tumours at different stages of carcinogenesis in order to accurately target PGE2-EP receptor signalling as a therapeutic anti-CRC strategy.

Materials and methods Reagents and antibodies PGE2, PGE1-OH and butaprost were obtained from Cayman Chemical Co. (Ann Arbor, MI, USA) and dissolved in either DMSO or ethanol (minimum 1 mM). The EP4 receptor antagonist ONO-AE3-208 (10 mM in DMSO) was a gift from ONO Pharmaceutical Co., Osaka, Japan. SC236 (20 mM in DMSO) was a gift from Pfizer Inc., New York, NY, USA. Indomethacin (100 mM in DMSO) was purchased from Sigma Chemical Co., Poole, UK. Rolipram (50 mM in DMSO), H89, PD98059, U0126 (all 10 mM in DMSO), forskolin (20 mM in DMSO) and ouabain (10 mM in methanol) were purchased from Calbiochem (Nottingham, UK). Blasticidin (5 mg/ml in distilled water) was from Invitrogen, Paisley, UK. Recombinant human epidermal growth factor (EGF) was obtained from R&D Systems Ltd, Abingdon, UK. Antibodies to pERK1 and 2 (Thr202/Tyr204), total ERK1/2, pCREB (Ser133) and total CREB, pPKB/AKT (Ser473) and total PKB/AKT, and PY were obtained from Cell Signalling Technology Inc. (Beverly, MA, USA). Anti-EGFR antibody (clone 1005) was obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). Antibodies to ZO-1 and occludin were obtained from Zymed (San Francisco, CA, USA). Antibody to human b-actin was obtained from Sigma Chemical Co. Antibody to the V5 epitope was purchased from Invitrogen. Horseradish peroxidase-conjugated secondary antibodies were obtained from DAKOcytomation Ltd (Ely, UK). Alexa Fluor 488- and 594-conjugated secondary antibodies were obtained from Molecular Probes, Paisley, UK (Invitrogen). Cell culture COS-7 cells (a gift from Dr Helen Ardley, University of Leeds) and 293T cells (a gift from Dr Gina Scott, University of Leeds) were cultured in Dulbecco’s modified Eagle medium (Invitrogen) with 10% (v/v) foetal bovine serum (FBS). HT-29 cells (ECACC, Porton Down, UK) were cultured in Rosewell’s Park Memorial Institute 1640 medium containing Glutamax


3017 supplemented with 10% (v/v) FBS, 1000 U/ml penicillin and 500 U/ml streptomycin (all Invitrogen). Viable cell counts were obtained using a haemocytometer and exclusion of 0.04% (v/v) Trypan blue (Sigma). Overexpression of the human EP4 receptor Human (h) EP4 receptor cDNA (Accession number L28175, nucleotides 370–1877) was obtained from pCEP-EP4 (a gift from J Regan, Tucson, AZ, USA Fujino et al., 2002) by PCR using PfuTurbo DNA polymerase (Stratagene, La Jolla, CA, USA) (951C for 5 min, 951C for 45 s, 551C for 45 s, 721C for 2 min for 35 cycles, then 721C for 10 min, with forward, 50 -CAGTGTGCTGGAATTCTGC-30 and reverse, 50 -TATA CATTTTTCTGATAAGTTCAGT-30 primers) and then inserted (nucleotides 370–1851) into the pTracer-EF/Bsd vector (Invitrogen) in-frame with a C-terminal V5 epitope-polyhistidine tag using EcoRI and EcoRV. DNA sequencing confirmed high fidelity PCR cloning and correct gene orientation. Transient transfection of COS-7 and 293T cells was achieved using 1 mg pTracer-EP4 and Genejuice (Novagen, Madison, WI, USA) in the presence of 2.5% (v/v) FBS overnight at 371C. Transfection efficiency was monitored by fluorescence microscopy for cycle 3-green fluorescent protein. The same protocol was used for production of stable HT-29 cell clones, followed by dilution cloning in selection medium containing 10 mg/ml blasticidin. Dual-luciferase assay of the b-catenin/ TCF reporter pTOPflash was performed as described (Gardner et al., 2004). Electron microscopy HT-29 cells on glass coverslips were fixed in 3% (w/v) glutaldehyde in 0.1 M phosphate buffer overnight at 41C. Following a wash with phosphate buffer for 10 min, coverslips were treated with 2% (w/v) osmium tetroxide in phosphate buffer for 1 h with mixing. After passing through a graded alcohol series, coverslips were embedded in resin (Agar Scientific Ltd, Stansted, UK) and incubated overnight at 701C. Embedded cells were removed from the coverslip with a razor blade and re-embedded perpendicularly in fresh resin as above. Transmission electron microscopy on ultrathin sections was performed with a Jeol 100-S electron microscope. Quantification of the cyst area of HT-29 cells Phase-contrast photomicrographs of HT-29 cells in 35 mm wells were obtained using an Olympus CKX41 microscope. The total cyst area in four quadrant images from each well was calculated using Lucia G software (Nikon UK Ltd, Kingston-upon-Thames, UK). Immunoblot analysis and immunoprecipitation Cells were lysed in ice-cold 50 mM Tris-HCl (pH 7.2). containing 0.137 M sodium chloride, 1% (v/v) Brij 96, 1 mM ethylenediaminetetra-acetic acid, 200 mM 4-(2-aminoethyl)benzenesulphonyl fluoride, 20 mM leupeptin and 1 mM pepstatin (all Sigma) and then passed through a QIAshredder homogenizer (QIAGEN Ltd, Crawley, UK). For phosphoprotein analysis, cells were cultured in serum-free medium for 24 h and lysed in PhosphoSafet extraction buffer (Novagen). Immunoblot analysis was performed as described (Gardner et al., 2004), except that membranes were blocked with 5% (w/v) dried skimmed milk in phosphate-buffered saline (PBS) for 1 h at 201C. Membranes were probed with primary antibodies (1:1000 dilution except b-actin (1/5000)) in PBS containing 3% (w/v) dried skimmed milk for 1 h at 201C or in PBS in the presence of 5% (w/v) bovine serum albumin and 0.1% (v/v) Tween-20 (both Sigma) for 16 h at 41C. Horseradish

peroxidase-conjugated secondary antibodies were diluted 1/ 2000 in PBS plus 5% (w/v) dried skimmed milk powder and incubated with blots for 1 h at 201C. Immunoreactivity was visualized using ECL chemiluminescence (Pierce, Chester, UK). For immunoprecipitation, cells were lysed in radioimmunoprecipitation assay buffer (Cell Signalling) containing phosphatase inhibitor cocktail 2 (Sigma) and 1 mM sodium orthovanadate, at 41C. EGF receptor was pulled down using 4 mg/ml anti-EGFR antibody for 24 h at 41C and then protein G-sepharose (Amersham Pharmacia, Little Chalfont, UK) for 18 h at 41C. Immunodetection of PY and EGFR was performed as above. pEGFR (Tyr1173) and total EGFR expression was measured by a cell-based immunoassay (FACE; Active Motif, Carlsbad, CA, USA). All EGFR experiments were performed on cells that had undergone serum starvation for 24 h beforehand. Indirect immunofluorescence Cells on glass coverslips were fixed in 100% methanol at 201C for 10 min and washed twice with PBS before incubation with primary antibody (V5, 1/200; ZO-1, 1/50; occludin, 1/25) in PBS containing 1% (w/v) dried skimmed milk, 0.1%. (v/v) Tween-20 for 2 h at 201C. Secondary antibodies (1/400 dilution) were incubated with coverslips in 1% (w/v) dried skimmed milk powder in PBS for 1 h at 201C. Cells were then treated with 0.2 mg/ml 40 , 6 diamidino-2phenylindole (DAPI) in PBS for 10 min followed by PBS washes (3 5 min). Coverslips were mounted in MOWIOL (Calbiochem). Assay of intracellular cAMP content HT-29 cells (1 106) were incubated overnight in triplicate 35 mm wells. Cells were pretreated with 50 mM rolipram for 10 min before addition of 10 mM ONO-AE3-208, followed 45 min later by EP receptor agonists. COX inhibitors were added 30 min before cAMP assay. Intracellular cAMP content was assayed using a Biotrak cAMP immunoassay (Amersham Pharmacia; non-acetylation protocol). Immunoassay of PGE2 levels in cell-conditioned medium HT-29-con and HT-29-EP4 cells (2.5 105). were seeded in triplicate in 24-well plates. At 24 h, fresh medium (0.5 ml) was replaced with or without PGE1-OH (10 nM or 1 mM) for 4 h before further replacement by fresh medium. Cell-free conditioned medium was collected after 24 h and adherent, viable cells were counted. PGE2 levels were measured by a highsensitivity immunoassay (R&D Systems Ltd). Based on the minimum detectable dose of PGE2 (10.1 pg/ml), the assay could detect PGE2 in cell-conditioned medium above a level of 2 pg/ml/105 cells. Measurement of cell proliferation and apoptosis HT-29-con and HT-29-EP4 cells (2 106) were seeded in triplicate 35 mm wells and incubated overnight. Medium containing ONO-AE3-208 or carrier alone was replaced 60 min before addition of PGE2 or carrier alone to the cultures. At 24 h, non-adherent cells were harvested by centrifugation at 800 g for 5 min and cells resuspended in 4% (w/v) paraformaldehyde in PBS and stored at 41C. Fluorescence microscopy of a minimum 100 DAPI-stained nonadherent cells was performed as described (Ko et al., 2002). Apoptosis was also measured in combined non-adherent and adherent cells (detached with Accutase (Sigma)) using an Annexin V:FITC Apoptosis Detection Kit (BD Pharmingen, San Diego, CA, USA). For these experiments, 5 105 cells Oncogene


3018 were seeded into 35 mm wells and analysed at 24 h using a BD LSR II Flow Cytometer (BD Biosciences, San Jose, CA, USA). Adherent cell counting experiments were performed by plating 5 104 cells in triplicate 35 mm wells and counting trypsinized, Trypan blue-negative cells at 24, 48 and 72 h. BrdU incorporation assays (Roche Diagnostics Ltd, Lewes, UK) were performed on 1 103 cells per well in 96-well plates between 24 and 72 h. In both assays, fresh medium containing PGE2 and/or ONO-AE3-208 was replaced at 48 h. Assay of anchorage-independent growth capability HT-29-con or HT-29-EP4 cells (5 104) were cultured in 0.4%. (w/v) agar as described (Ko et al., 2002). Fresh medium containing working concentrations of PGE2 and/or ONO-AE3-208 was added every 48 h. On day 14, colonies greater than 200 mm in diameter were counted. The % colony formation efficiency was calculated as the (number of colonies/5 104) 100. Real-time PCR of EP4 receptor mRNA levels in human colorectal tissue One microlitre of each cDNA from the human colon matched cDNA pair panel (Clontech, Mountain View, CA, USA) was

amplified by SYBR Green real-time PCR using an ABI Prism 7700 Sequence Detection System (Applied Biosystems, Warrington, UK). Primers (0.3 pmol each) for human EP4 receptor cDNA were; forward dTCTTACTCATTGCCACCTCCCT and reverse dCTTGGCTGATATAACTGGTTGACG. Reactions were incubated at 501C for 2 min and then 951C for 10 min, followed by 50 cycles consisting of 951C for 15 s and 601C for 1 min. Differences in transcript levels between paired tumour/normal samples were determined by the comparative 2DCt method.

Acknowledgements We thank Dr John Regan for the kind gift of the pCEP4-EP4 plasmid. Dr Ewan Morrison gave valuable help with timelapse microscopy. Dr Valerie Speirs provided human EP4 receptor primers for real-time PCR. This work was funded by the Association for International Cancer Research. MA Hull is a Medical Research Council (UK) Senior Clinical Fellow, work in his laboratory is also funded by the World Cancer Research Fund and Yorkshire Cancer Research.

References Adamson RH, Liu B, Nilson Fry G, Rubin LL, Curry FE. (1998). Microvascular permeability and number of tight junctions are modulated by cAMP. Am J Physiol Heart Circ Physiol 274: H1885–H1894. Breyer RM, Bagdassarian CK, Myers SA, Breyer MD. (2001). Prostanoid receptors: subtypes and signalling. Annu Rev Pharmacol Toxicol 41: 661–690. Buchanan FG, Gorden DL, Matta P, Shi Q, Matrisian LM, DuBois RN. (2006). Role of b-arrestin 1 in the metastatic progression of colorectal cancer. Proc Natl Acad Sci USA 103: 1492–1497. Buchanan FG, Wang D, Bargiacchi F, DuBois RN. (2003). Prostaglandin E2 regulates cell migration via the intracellular activation of the epidermal growth factor receptor. J Biol Chem 278: 35451–35457. Cahlin C, Gelin J, Andersson M, Lo¨nroth C, Lundholm K. (2005). The effects of non-selective, preferential-selective and selective COX-inhibitors on the growth of experimental and human tumors in mice related to prostanoid receptors. Int J Oncol 27: 913–923. Cassano G, Gasparre G, Susca F, Lippe C, Guanti G. (2000). Effect of prostaglandin E2 on proliferation, Ca2+ mobilization and cAMP in HT-29 human colon adenocarcinoma cells. Cancer Lett 152: 217–222. Castellone MD, Teramoto H, Williams BO, Druey KM, Gutkind JS. (2005). Prostaglandin E2 promotes colon cancer growth through a novel Gs-axin-b-catenin signalling axis. Science 310: 1504–1510. Chell SD, Witherden IR, Dobson RR, Moorghen M, Herman AA, Qualtrough D et al. (2006). Increased EP4 receptor expression in colorectal cancer progression promotes cell growth and anchorage independence. Cancer Res 66: 3106–3113. Colucci R, Blandizzi C, Tanini M, Vassalle C, Breschi MC, Del Tacca M. (2005). Gastrin promotes human colon cancer cell growth via CCK-2 receptor-mediated cyclooxygenase-2 induction and prostaglandin E2 production. Br J Pharmacol 144: 338–348. Desai S, Ashby B. (2001). Agonist-induced internalization and mitogen-activated protein kinase activation of the human prostaglandin EP4 receptor. FEBS Lett 501: 156–160. Oncogene

Dye JF, Leach L, Clark P, Firth JA. (2001). Cyclic AMP and acidic fibroblast growth factor have opposing effects on tight and adherens junctions in microvascular endothelial cells in vitro. Microvasc Res 62: 94–113. Elder DJE, Halton DE, Hague A, Paraskeva C. (1997). Induction of apoptotic cell death in human colorectal carcinoma cell lines by a cyclooxygenase-2 (COX-2)-selective nonsteroidal anti-inflammatory drug: Independence from COX-2 protein expression. Clin Cancer Res 3: 1679–1683. Fujino H, Regan JW. (2006). EP4 prostanoid receptor coupling to a pertussis toxin sensitive inhibitory G-protein. Mol Pharmacol 69: 13–18. Fujino H, Salvi S, Regan RW. (2005). Differential regulation of phosphorylation of the cAMP response element-binding protein after activation of EP2 and EP4 prostanoid receptors by prostaglandin E2. Mol Pharmacol 68: 251–259. Fujino H, West KA, Regan JW. (2002). Phosphorylation of glycogen synthase kinase-3 and stimulation of T-cell factor signalling following activation of EP2 and EP4 prostanoid receptors by prostaglandin E2. J Biol Chem 277: 2614–2619. Fujino H, Xu W, Regan RW. (2003). Prostaglandin E2 induced functional expression of early growth response factor-1 by EP4, but not EP2, prostanoid receptors via the phosphatidylinositol 3-kinase and extracellular signalregulated kinases. J Biol Chem 278: 12151–12156. Gardner SH, Hawcroft G, Hull MA. (2004). The effect of nonsteroidal anti-inflammatory drugs on b-catenin protein levels and catenin-related transcription in human colorectal cancer cells. Br J Cancer 91: 153–163. Grosser T, Fries S, FitzGerald GA. (2006). Biological basis for the cardiovascular consequences of COX-2 inhibition: therapeutic challenges and opportunities. J Clin Invest 116: 4–15. Han C, Wu T. (2005). 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. J Biol Chem 280: 24053–24063. Han C, Michalopoulos GK, Wu T. (2006). Prostaglandin E2 receptor EP1 transactivates EGFR/MET receptor tyrosine


3019 kinases and enhances invasiveness in human hepatocellular carcinoma cells. J Cell Physiol 207: 261–270. Holla VR, Mann JR, Shi Q, DuBois RN. (2005). Prostaglandin E2 regulates the nuclear receptor NR4A2 in colorectal cancer. J Biol Chem 281: 2676–2682. Hoshino T, Tsutsumi S, Hwang H-J, Tsuchiya T, Mizushima T. (2003). Prostaglandin E2 protects gastric mucosal cells from apoptosis via EP2 and EP4 receptor activation. J Biol Chem 278: 12752–12758. Hsi LC, Baek SJ, Eling TE. (2000). Lack of cyclooxygenase-2 activity in HT-29 human colorectal carcinoma cells. Exp Cell Res 256: 563–570. Hull MA, Ko SCW, Hawcroft G. (2004). Prostaglandin EP receptors: Targets for treatment and prevention of colorectal cancer? Mol Cancer Ther 3: 1031–1039. Hull MA. (2005). Cyclooxygenase-2: How good is it as a target for cancer chemoprevention? Eur J Cancer 41: 1854–1863. Kato S, Aihara E, Yoshii K, Takeuchi K. (2005). Dual action of prostaglandin E2 on gastric acid secretion through different EP-receptor subtypes in the rat. Am J Physiol Gastrointest Liver Physiol 289: G64–G69. Keese M, Magdeburg RJ, Herzog T, Hasenberg T, Offterdinger M, Pepperkok R et al. (2005). Imaging epidermal growth factor receptor phosphorylation in human colorectal cancer cells and human tissues. J Biol Chem 280: 27826–27831. Kiriyama M, Ushikubi F, Kobayashi T, Hirata M, Sugimoto Y, Narumiya S. (1997). Ligand binding specificities of the eight types and subtypes of the mouse prostanoid receptors expressed in chinese hamster ovary cells. Br J Pharmacol 122: 217–224. Ko SCW, Chapple KS, Hawcroft G, Coletta PL, Markham AF, Hull MA. (2002). Paracrine cyclooxygenase-2-mediated signalling by macrophages promotes tumorigenic progression of intestinal epithelial cells. Oncogene 21: 7175–7186. Krause W, DuBois RN. (2001). Eicosanoids and the large intestine. Prostagl and Other Lipid Mediator 61: 145–161. Ma X, Kundu N, Rifat S, Walser T, Fulton AM. (2006). Prostaglandin E receptor EP4 antagonism inhibits breast cancer metastasis. Cancer Res 66: 2923–2927. Mayr B, Montminy M. (2001). Transcriptional regulation by the phosphorylation-dependent factor CREB. Nature Rev Mol Cell Biol 2: 599–609. Mendez M, LaPointe MC. (2005). PGE2-induced hypertrophy of cardiac myocytes involves EP4 receptor-dependent activation of p42/44 MAPK and EGFR transactivation. Am J Physiol Heart Circ Physiol 288: H2111–H2117. Misra UK, Pizzo SV. (2005). Coordinate regulation of forskolin-induced cellular proliferation in macrophages by protein kinase A/cAMP-response element-binding protein (CREB) and Epac1-Rap1 signaling. J Biol Chem 280: 38276–38289. Mutoh M, Watanabe K, Kitamura T, Shoji Y, Takahashi M, Kawamori T et al. (2002). Involvement of prostaglandin E receptor subtype EP4 in colon carcinogenesis. Cancer Res 62: 28–32. Nishihara H, Hwang M, Kizaka-Kondoh S, Eckmann L, Insel PA. (2004). Cyclic AMP promotes cAMP-responsive element-binding protein-dependent induction of cellular inhibitor of apoptosis protein-2 and suppresses apoptosis

of colon cancer cells through ERK1/2 and p38 MAPK. J Biol Chem 279: 26176–26183. Nishihara H, Kizaka-Kondoh S, Insel PA, Eckmann L. (2003). Inhibition of apoptosis in normal and transformed intestinal epithelial cells by cAMP through induction of inhibitor of apoptosis protein (IAP)-2. Proc Natl Acad Sci USA 100: 8921–8926. Ophir I, Cohen E, Ben Shaul Y. (1995). Apical polarity in human colon carcinoma cell lines. Tissue Cell 27: 659–666. Pai R, Soreghan B, Szabo IL, Pavelka M, Baatar D, Tarnawski AS. (2002). Prostaglandin E2 transactivates EGF receptor: a novel mechanism for promoting colon cancer growth and gastrointestinal hypertrophy. Nature Med 8: 289–293. Pavlovic S, Du B, Sakamoto K, Khan KMF, Natarajan C, Breyer RM et al. (2006). Targeting PGE2 receptors as an alternative strategy to block COX-2-dependent extracellular matrix-induced MMP-9 expression by macrophages. J Biol Chem 281: 3321–3328. Pozzi A, Yan X, Macias-Perez I, Wei S, Hata AN, Breyer RM et al. (2004). Colon carcinoma cell growth is associated with prostaglandin E2-EP4 receptor-evoked ERK activation. J Biol Chem 279: 29797–29804. Regan RW. (2003). EP2 and EP4 prostanoid receptor signalling. Life Sci 74: 143–153. Sales KT, Maudsley S, Jabbour HN. (2004). Elevated prostaglandin EP2 receptor in endometrial adenocarcinoma cells promotes vascular endothelial growth factor expression via cyclic 30 , 50 -adenosine monophosphate-mediated transactivation of the epidermal growth factor receptor and extracellular signal-regulated kinase 1/2 signalling pathways. Mol Endocrinol 18: 1533–1545. Seifert R, Wenzel-Seifert K. (2002). Constitutive activity of G-protein-coupled receptors: cause of disease and common property of wild-type receptors. Naunyn-Schmiedeberg’s Arch Pharmacol 366: 381–416. Shao J, Sheng H, Inoue H, Morrow JD, DuBois RN. (2000). Regulation of constitutive cyclooxygenase-2 expression in colon carcinoma cells. J Biol Chem 275: 33951–33956. Sheng H, Shao J, Washington MK, DuBois RN. (2001). Prostaglandin E2 increases growth and motility of colorectal carcinoma cells. J Biol Chem 276: 18075–18081. Shoji Y, Takahashi M, Kitamura T, Watanabe K, Kawamori T, Maruyama T et al. (2004). Downregulation of prostaglandin E receptor subtype EP3 during colon cancer development. Gut 53: 1151–1158. Smith M-L, Hawcroft G, Hull MA. (2000). The effect of nonsteroidal anti-inflammatory drugs on human colorectal cancer cells: evidence of different mechanisms of action. Eur J Cancer 36: 664–674. Wang D, Buchanan FG, Wang H, Dey SK, DuBois RN. (2005). Prostaglandin E2 enhances intestinal adenoma growth via activation of the Ras-mitogen-activated protein kinase cascade. Cancer Res 65: 1822–1829. Wang D, Wang H, Shi Q, Katkuri S, Walhi W, Desvergne B et al. (2004). Prostaglandin E2 promotes colorectal adenoma growth via transactivation of the nuclear peroxisome proliferator-activated receptor d. Cancer Cell 6: 285–295. Wright CW, Duckett CS. (2005). Reawakening the cellular death program in neoplasia through therapeutic blockade of IAP function. J Clin Invest 115: 2673–2678.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene