Oestrogen receptor-α contributes to the regulation of the hedgehog signalling pathway in ERα-positive gastric cancer

Background: Oestrogen receptor-alpha (ERα) is highly expressed in diffuse-type gastric cancer and oestrogen increases the proliferation of ERα-positive gastric cancer. However, a detailed mechanism by which oestrogen increases the proliferation of these cells is still unclear. Methods: We used 17-β-oestradiol (E2) as a stimulator against the ERα pathway. Pure anti-oestrogen drug ICI 182 780 (ICI) and small interfering RNA against ERα (ERα siRNA) were used as inhibitors. Cyclopamine (Cyc) was used as the hedgehog (Hh) pathway inhibitor. Two human ERα-positive gastric cancer cells were used as target cells. Effects of the stimulator and inhibitor on E2-induced cell proliferation were also examined. Results: In ERα-positive cells, E2 increased not only cell proliferation but also one of the ligands of the Hh pathway, Shh expression. 17-β-Oestradiol-induced cell proliferation was suppressed by ICI, ERα siRNA or Cyc. The increased expression of Shh induced by E2 was suppressed by ICI and ERα siRNA but not by Cyc. Furthermore, recombinant Shh activated the Hh pathway and increased cell proliferation, whereas anti-Shh antibody suppressed E2-induced cell proliferation. When a relationship between ERα and Shh expressions was analysed using surgically resected gastric cancer specimens, a positive correlation was found, suggesting a linkage between the ERα and Hh pathways. Conclusion: Our data indicate that activation of the ERα pathway promotes cell proliferation by activating the Hh pathway in a ligand-dependent manner through Shh induction of ERα-positive gastric cancer.

Gastric cancer is one of the most lethal of all malignancies (Parkin, 2001). In particular, patients with unresectable, metastatic or recurrent gastric cancer have only few therapeutic options. Thus, it is urgent to develop novel therapeutics for gastric cancer.
Gastric cancers are classified into two major histological types using the criteria of Lauren, namely intestinal type and diffuse type (Lauren, 1965). The carcinogenic pathway of intestinal-type gastric cancer develops through several sequential stages with Helicobacter pylori infection, followed by chronic gastritis, atrophic gastritis and intestinal metaplasia (Yuasa, 2003), whereas the diffuse type is believed to be derived from hyperplastic stem or precursor cells (Ming, 1998). It is considered that diffuse-type gastric cancer has a higher malignant potential such as invasion and metastasis, compared with intestinal-type gastric cancer.
Oestrogen has various physiological functions, such as normal cell growth and differentiation in many target tissues. Oestrogen is produced not only from the ovary but also from extra-ovarian tissues, that is, from the skin, brain, testis, adipose tissues and vascular smooth muscle (Ackerman et al, 1981;Leshin et al, 1981;Brodie and Inkster, 1993;Lephart, 1996;Ueyama et al, 2002). The biological actions of oestrogen are mediated through two specific receptors (oestrogen receptor, ERs), ERa and ERb, which belong to the nuclear receptor superfamily. Oestrogen-bound ERs bind as homodimers or heterodimers to a specific DNA sequence known as the oestrogen-responsive element (ERE), and regulate the transcription of target genes (Matthews and Gustafsson, 2003). As oestrogen is a stimulant for the initiation and promotion of breast cancer, and ERa can be expressed in gastric cancer cells, it has been suggested that the ERa pathway may have a role in the progression of gastric cancer (Harrison et al, 1989b;Wu et al, 1994;Karat et al, 1999;Takano et al, 2002;Chandanos et al, 2008). In gastric cancer, ERa expression is higher in the diffuse type than in the intestinal type (Kitaoka, 1983;Tokunaga et al, 1986;Matsui et al, 1992;Zhao et al, 2003), and oestradiol (E2) enhances gastric cancer cell proliferation in vitro (Matsui et al, 1992;Takano et al, 2002). On the basis of these observations, several clinical trials using a partial oestrogen antagonist, tamoxifen (TAM), have been conducted for the management of ERa-positive gastric cancer patients. However, the results have not been consistent and the utility of ERa for the treatment of gastric cancer is still controversial (Harrison et al, 1989a, b;Koullias et al, 2003).
The hedgehog (Hh) signalling pathway has a crucial role in embryonic development, tissue regeneration and carcinogenesis. Of the three Hh ligands, namely Sonic Hh (Shh), Indian Hh (Ihh) and Desert Hh (Dhh) (Pasca di Lum and Beachy, 2004;Briscoe and Therond, 2005;Hooper and Scott, 2005;Katoh and Katoh, 2005), Shh is exclusively expressed in the acidproducing parietal cells in both human and murine stomach, and is believed to be a regulator of gastric fundic gland differentiation and mutation (Ramalho-Santos et al, 2000;van den Brink et al, 2001van den Brink et al, , 2002Dimmler et al, 2003). Pharmacological inhibition of Shh signalling in the adult stomach causes loss of parietal cells (gastric atrophy) and subsequent disruption of glandular differentiation (van den Brink et al, 2002). Furthermore, Shh is strongly implicated in maintaining stem cell niches in the adult stomach (Katoh and Katoh, 2006). Recently, ligand-dependent activation of the Hh pathway, especially due to an aberrant expression of Shh, has been detected in various cancers including gastric cancer Watkins et al, 2003;Karhadkar et al, 2004;Sanchez et al, 2004;Sheng et al, 2004). Overexpression of Shh leads to carcinogenesis through the aberrant activation of the Hh pathway Thayer et al, 2003;Cengel, 2004;Sanchez et al, 2004). It is also believed that Shh has an important role during the progression of gastric cancer, because Shh expression is much higher in the diffuse type than in the intestinal type in human gastric cancer (Ma et al, 2005(Ma et al, , 2006. It is noteworthy that our previous study demonstrated a positive correlation between Hh pathway activation and ERa status (Kubo et al, 2004), and suggested that ERa could regulate Hh pathway activation in ERa-positive breast cancer . It has also been shown that the Hh pathway can be a useful therapeutic target against breast cancer and gastric cancer (Kubo et al, 2004;Yanai et al, 2007). From these results, we hypothesised that stimulation of the ERa pathway induces Shh expression, activates the Hh pathway and consequently promotes cell proliferation in ERa-positive gastric cancer cells. The ERa pathway could be a possible therapeutic target for patients with ERa-positive gastric cancer.

Expression vectors
cDNAs encoding human ERa were cloned into the pSG5 expression vector as described previously (Green et al, 1988;Kato et al, 2002). This cDNA and ERE-tk-luc, a single consensus ERE upstream of luciferase, were kindly provided by Dr Norio Wake (Kyushu University, Fukuoka, Japan). pRL-SV40 was purchased from Promega (Madison, WI, USA).

Immunoblotting
Whole-cell extraction was performed with M-PER Reagents (Pierce Biotechnology, Rockford, IL, USA) according to the manufacturer's instructions. Protein concentration was determined using the Bio-Rad Protein Assay (Bio-Rad Laboratories), and whole-cell extract (80 mg) was separated by electrophoresis on a SDSpolyacrylamide gel, and transferred to Protran nitrocellulose membranes (Whatman, Dassel, Germany). Blots were then incubated overnight with GAPDH (1 : 1000), ERa (1 : 200) or Shh (1 : 200) primary antibody at 41C. Blots were then incubated in HRP-linked secondary antibody (Amersham Biosciences, Piscataway, NJ, USA) at room temperature for 1 h. Immunocomplexes were detected using ECL together with the western blotting detection system (Amersham Biosciences) and visualised using a Molecular Imager FX (Bio-Rad Laboratories). Glyceraldehyde-3phosphate dehydrogenase was used as a protein loading control.

Dual luciferase assay
KATOIII and NCI-N87 cells in 24-well plates were transfected with plasmids with TransFast transfection reagent according to the manufacturer's instructions. Cells on each well were co-transfected with 10 ng of pRL-SV40 (Promega) and 1 mg ERE-tk-Luc. After oestrogen starvation, E2 and ICI or Cyc were added to each well for 8 h, and luciferase assays were performed using the dual luciferase assay kit (Promega) according to the manufacturer's instructions. The luciferase activities were normalised to Renilla luciferase activity.

Small interfering RNA against ERa
KATOIII and NCI-N87 cells (1.0 Â 10 6 cells) were transfected with small interfering RNA (siRNA) (100 nM) against ERa by lipofectamine as per the manufacturer's instructions, and then plated in a 25-cm 2 flask for 24 h in 10% FBS-RPMI. After oestrogen starvation, the cells were treated with E2 for 16 h, and then used for real-time RT -PCR. The following siRNAs were used: Validated Stealth RNAi against ERa and the Stealth RNAi-negative control (Invitrogen).

Proliferation assay
KATOIII (5 Â 10 3 per well), NCI-N87 (1 Â 10 4 per well) and MK-1 (5 Â 10 3 per well) cells were seeded in 48-well plates in complete culture medium and were incubated overnight. After oestrogen starvation, the medium was changed to 5% DCC-FBS-MEM containing various concentrations of reagents. After 72 h of incubation, cells were harvested by trypsinisation, and viable cells were counted using a Coulter counter (Beckman Coulter, Fullerton, CA, USA).

Clinical samples
Surgical specimens were obtained from 20 patients with diffusetype gastric cancer and from 20 patients with intestinal type. All of the patients underwent resection at the Department of Surgery and Oncology, Kyushu University (Fukuoka, Japan), between 1996 and 2004. All 40 patients gave informed consent before surgical treatment and were enrolled into this study. All surgical specimens were frozen at À801C, examined histopathologically and classified using the tumour-node-metastasis classification. Total mRNA of these specimens was extracted using the RNeasy mini kit (Qiagen) as per the manufacturer's recommendation.

Immunohistochemistry
Single-antibody detection was carried out as described previously (Kubo et al, 2004), with the following protocol modifications: Endogenous peroxidase activity was blocked using 3% H 2 O 2 in methanol for 30 min at room temperature. Antigen retrieval was achieved by boiling tissue in 0.01 mol l À1 sodium citrate (pH 6.0) for 5 min. All primary antibodies were incubated overnight at 41C. The primary antibodies used were Shh (1 : 50 H-160, sc-9024,) and ERa (1 : 50 F-10, sc-8002) (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Secondary antibodies (Shh rabbit anti-IgG; ERa, mouse anti-IgG; Nichirei Co., Ltd, Tokyo, Japan) were applied for 1 h at room temperature. Protein was detected by brown pigmentation using the standard 3, 3 0 -diaminobenzidine (DAB) protocol. Slides were lightly counterstained with haematoxylin. Negative controls were obtained in all cases by omitting the first antibodies. All primary antibodies had been previously tested for immunohistostaining.

Statistical analysis
Student's t-test was used for statistical analysis, unless indicated otherwise. The correlation of clinicopathological features between the ERa and Shh expression was analysed by the Mann -Whitney U-test. All calculations were carried out using the StatView 5.0 J software (Abacus Concepts, Berkeley, CA, USA). Single asterisk denotes Po0.05; P-values o0.05 were considered significant.

Oestrogen activates the ERa pathway and induces increased proliferation
We examined ERa expression in eight human gastric cancer cell lines. Five cell lines showed ERa expression at the mRNA level ( Figure 1A), and three of the five cell lines, namely KATOIII, NCI-N87 and AGS cells, also showed detectable ERa expression at the protein level ( Figure 1A). Thus, we selected KATOIII (diffuse type) and NCI-N87 (intestinal type) cells as ERa-positive cells, and selected MK-1 cells (diffuse type) as ERa-negative cells. We then examined whether the ERa pathway has a role in the proliferation of ERa-positive cells. As we expected, E2 increased the proliferation of ERa-positive but not negative cells in a dose-dependent manner ( Figure 1B). This E2-induced cell proliferation was completely inhibited by ICI 182 780 (100 nM). ICI did not affect cell proliferation without supplementation of E2 ( Figure 1B). On the basis of these preliminary data, E2 and ICI were used at a concentration of 3 and 100 nM, respectively, throughout the study. We further examined whether the ERa pathway is functional in these ERa-positive cells. Activation of the ERa pathway by E2 was determined with an ERa reporter assay and pS2 expression, a target gene of the ERa pathway (Campbell- Thompson, 1997;Singh et al, 1997). We found that E2 increased both ERa-responsive reporter activity ( Figure 1C) and pS2 mRNA expression ( Figure 1C) in these ERa-positive cells. This increased activation of the ERa pathway by E2 was almost completely inhibited by ICI. Finally, to confirm the contribution of the ERa pathway to E2-induced proliferation of ERa-positive cells, we silenced ERa mRNA expression of ERa-positive cells by RNA interference. Transfection of siRNA resulted in an 85% or greater knockdown of ERa mRNA expression (data not shown). Silencing of ERa did not affect cell proliferation without supplementation of E2. However, E2-induced cell proliferation was inhibited by ERa-siRNA ( Figure 1D). These data indicate that E2 increases proliferation of ERa-positive gastric cancer cells through activation of the ERa pathway.

Oestrogen-induced cell proliferation is suppressed by inhibition of the Hh pathway
As our previous study indicated a crosstalk between the ERa and Hh pathways in ERa-positive breast cancer cells , we examined the expression of Hh-related molecules, including Shh, Ptched 1 (Ptc) and Gli1, in ERa-positive gastric cancer cells. Consistent with our previous report (Yanai et al, 2007), we found that these cells expressed Hh-related molecules at the mRNA level ( Figure 2A). Gli1 expression indicates a constitutive activation of the Hh pathway in gastric cancer cells, because Gli1 is a target gene of the Hh pathway (Kubo et al, 2004). In our previous report, cyclopamine suppressed the proliferation of these cells (Yanai et al, 2007). To estimate a contribution of the Hh pathway to E2-induced increase in proliferation, we compared the increasing rate of proliferation induced by E2 between the presence and absence of Hh pathway activity ( Figure 2B). In Kato   increasing rate was 28 and 18% in the presence and absence of Hh pathway activity, respectively. In NCI-N87, it was 35 and 22%, respectively. These data indicate that a block of the Hh pathway reduces the degree of E2-induced increase of proliferation. To determine the relationship between these two pathways, we examined how cyclopamine affects ERa pathway activation induced by E2. We found that cyclopamine did not affect E2induced activation of the ERa pathway ( Figure 2C). These data suggest that the Hh pathway contributes to E2-induced cell proliferation of ERa-positive gastric cancer cells.

Oestrogen induces the increased Shh expression through activation of the ERa pathway
As contribution of the Hh pathway to E2-induced cell proliferation has been previously suggested, we examined whether E2 could induce Hh pathway activation in ERa-positive cells. Activation of the Hh pathway was determined by Gli1 mRNA expression. We found that E2 markedly increased Gli1 mRNA expression in ERapositive cells, and that the increased Gli1 mRNA expression was significantly suppressed by ICI. ICI alone did not affect Gli1 mRNA expression without supplementation of E2 ( Figure 3A). Cyclopamine suppressed this E2-induced Gli1 mRNA expression more strongly than did ICI. 17-b-Oestradiol-induced Gli1 mRNA expression was not found in ERa-negative cells. We then investigated the molecular mechanisms by which E2 activates the Hh pathway in ERa-positive cells. Ligand-dependent activation of the Hh pathway has been shown in gastric cancer . Therefore, we focused on an Hh pathway ligand, Shh. As expected, E2 increased Shh mRNA expression of ERa-positive cells, and this E2-induced increase was almost completely inhibited by ICI ( Figure 3B). Cyclopamine did not affect E2-induced Shh expression. Neither ICI nor cyclopamine significantly affected Shh expression in ERa-negative cells. Finally, to confirm a contribution of the ERa pathway to E2-induced Shh mRNA expression in ERapositive cells, we used ERa-silenced ERa-positive cells as target  cells. Silencing of ERa alone did not affect Shh expression when there was no supplementation of E2. Importantly, E2-induced Shh expression was almost completely inhibited in ERa-silencing cells ( Figure 3C). These data indicate that E2 increases Shh induction of ERa-positive gastric cancer cells through the activation of the ERa pathway.

Oestrogen increases cell proliferation through Shh induction, followed by increased Hh pathway activation
We next examined whether Shh induced by E2 could cause the proliferation of ERa-positive cells in a ligand-dependent manner.
As mentioned above, the Hh pathway is constitutively activated in these cells. Therefore, we first investigated whether exogenous Shh could further increase the proliferation of these cells. We found that rhShh increased Gli1 mRNA expression (data not shown) and the proliferation of these ERa-positive cells in a dose-dependent manner ( Figure 4A). We investigated whether a neutralising antibody against Shh could suppress E2-induced proliferation of ERa-positive cells. Anti-Shh antibody decreased Gli1 mRNA expression (data not shown). Anti-Shh antibody reduced the increasing rate of proliferation induced by E2 ( Figure 4B). In KATO III, the increasing rate was 2 and 33% in the presence and absence of the anti-Shh antibody, respectively. In NCI-N87, it was   36 and 42%, respectively. These data indicate that the anti-Shh antibody reduces the degree of E2-induced increase of proliferation, suggesting a contribution of Shh to E2-induced proliferation. Taken together, these results suggest that Shh induced by E2 is functional and increases cell proliferation by activating the Hh pathway in a ligand-dependent manner.
ERa mRNA expression positively correlates with Shh mRNA expression in gastric cancer tissues We investigated the possibility that a correlation between the ERa and Hh pathways exists in vivo using 40 surgically resected gastric cancer tissues. When 40 gastric cancer tissues were histologically divided into two groups, that is, 20 diffuse type and 20 intestinal type, no significant correlation between histological type and UICC stage was found (data not shown). The ERa mRNA expression was significantly higher in diffuse-type gastric cancer tissues than in intestinal-type gastric cancer tissues (Po0.0001; Figure 5A). Similarly, Shh mRNA expression was significantly higher in diffuse-type gastric cancer tissues than in intestinal-type gastric cancer tissues (Po0.016; Figure 5A). As expected, mRNA expression levels of ERa and Shh were tightly correlated (Po0.0001; Figure 5B). The correlation between ERa and Shh is significantly stronger in diffuse-type (Po0.0001; Supplementary Figure 1A) gastric cancer tissues than in intestinal-type gastric cancer tissues (Po0.0226; Supplementary Figure 1B). Expression of ERa and Shh was also examined immunohistochemically. Consistent with mRNA expression, both ERa and Shh were also highly expressed in diffuse-type gastric cancer cells ( Figure 5C). In addition, histochemical analysis showed that Shh is mainly expressed by cancer cells themselves. In this study, however, we did not perform correlation analysis between these molecules at the protein level, because it was difficult to accurately estimate the expression intensity. Nevertheless, our data suggest that a novel link between the ERa and Hh pathways is present even in gastric cancer tissues.

DISCUSSION
We have shown, for the first time, a biologically significant linkage between the ERa and Hh pathways in ERa-positive gastric cancer cells. Briefly, oestrogen activates the ERa pathway, induces Shh production, activates Hh activation in an Shh-dependent manner and consequently increases cell proliferation in ERa-positive gastric cancer cells. Our data suggest that the ERa pathway could be a possible therapeutic target for patients with ERa-positive gastric cancer.
Consistent with our present data, it has already been shown that both ERa (Kitaoka, 1983;Tokunaga et al, 1986;Matsui et al, 1992;Zhao et al, 2003) and Shh (Ma et al, 2005;Yanai et al, 2007) are frequently expressed in diffuse-type gastric cancer, compared with intestinal type. Recently, it was shown that the Hh pathway might be linked to other proliferation-related signalling pathways. We have shown a crosstalk of the Hh pathway with the nuclear factor k-B pathway in pancreatic cancer (Nakashima et al, 2006), the Wnt pathway in colonic and gastric cancer Yanai et al, 2008) and the ERa pathway in breast cancer . Thus, we speculated that there was a linkage between the ERa and Hh pathways in diffuse-type gastric cancer. Our present data showed two key points in the linkage between the two pathways. One finding is that the ERa pathway affects the Hh pathway, because E2 induces both ERa pathway activation and Shh induction, followed by Hh pathway activation, and all of these phenomena were almost completely suppressed by blockade of the ERa pathway. The other finding is that the Hh pathway does not affect the ERa pathway, because neither blockade nor stimulation of the Hh pathway affected ERa pathway activation. A key molecule linking the ERa pathway with the Hh pathway is Shh, which is induced by ERa pathway activation. Thus, we speculate that a crosstalk between the two pathways is a one-way link from the ERa to the Hh pathway. The molecular mechanisms by which ERa pathway activation induces Shh production remain unclear.
Ligand-dependent Hh pathway activation participates in the increased proliferation of gastric cancer Yanai et al, 2007). Therefore, in our study, it is not surprising that oestrogen-induced Shh induction increased the proliferation of ERa-positive gastric cancer cells. Our data indicate that blockade of the ERa pathway, such as using anti-oestrogens, may be valuable therapeutic tools for patients with ERa-positive gastric cancer. Interestingly, several clinical randomised controlled studies have already been performed using the selective ER modulator, TAM. Although ER status is an independent poor prognostic factor in gastric cancer (Harrison et al, 1989b;Matsui et al, 1992), the results of clinical randomised controlled studies disappointed us (Harrison et al, 1989a, b). Nevertheless, our present data still indicate a possibility of an ERa pathway blocker as a therapeutic tool for patients with ERa-positive gastric cancer. When TAM was used at concentrations (1 mM) that do not affect cell proliferation of ERa-negative MK-1 cells (Supplementary Figure 2B), TAM did not significantly suppress both E2-induced Hh activation and E2-induced cell proliferation of ERa-positive gastric cancer cells (Supplementary Figure 2A). In contrast, ICI, at a concentration (100 nM) that does not affect cell proliferation of MK-1cells, almost completely suppressed both E2-induced Hh activation and E2induced cell proliferation. It is well known that TAM can block only one domain, that is, activating function 2 (AF-2), among two distinct ERa activation domains (Ali and Coombes, 2002), whereas ICI can block both of the activation domains. In addition, it has been reported that TAM seems to function as a partial agonist in uterine tissues (Kedar et al, 1994). These findings indicate a possibility that different anti-oestrogens may display different tissue-specific actions. Our data indicate that the development of clinically available antagonists that can sufficiently block oestrogen-induced cell proliferation is promising for management of patients with aggressive diffuse-type gastric cancer. Our data also indicate that a combination of Hh inhibitors and anti-oestrogens may be more effective against ERa-positive gastric cancer. Furthermore, molecules contributing to E2-induced Shh induction may be novel therapeutic candidate molecules for the management of gastric cancer patients.  The representative photographs of immunohistochemistry for ERa and Shh in a diffuse-type gastric cancer tissue. ERa expression (arrows) was observed in gastric cancer cells (upper left panels). Shh expression (arrow heads) was also observed in gastric cancer cells (upper right panels). The photographs stained with secondary antibody alone are shown as control (lower panels).