Suppression of Wnt Signaling by the Green Tea Compound (–)-Epigallocatechin 3-Gallate (EGCG) in Invasive Breast Cancer Cells

Genetic and biochemical de-regulation of Wnt signaling is correlated with breast and other cancers. Our goal was to identify compounds that block Wnt signaling as a first step toward investigating new strategies for suppression of invasive and other breast cancers. In a limited phytonutrient screen, EGCG ((–)-epigallocatechin 3-gallate), the major phytochemical in green tea, emerged as an intriguing candidate. Epidemiological studies have associated green tea consumption with reduced recurrence of invasive and other breast cancers. Wnt signaling was inhibited by EGCG in a dose-dependent manner in breast cancer cells. The apparent mechanism targeted the HBP1 transcriptional repressor, which we had previously characterized as a suppressor of Wnt signaling. EGCG treatment induced HBP1 transcriptional repressor levels through an increase in HBP1 mRNA stability, but not transcriptional initiation. To test functionality, DNA-based short hairpin RNA (shRNA) was used to knockdown the endogenous HBP1 gene. Consistently, the HBP1 knockdown lines had reduced sensitivity to EGCG in the suppression of Wnt signaling and of a target gene (c-MYC). Because our ongoing studies clinically link abrogation of HBP1 with invasive breast cancer, we tested if EGCG also regulated biological functions associated with de-regulated Wnt signaling and with invasive breast cancer. EGCG reduced both breast cancer cell tumorigenic proliferation and invasiveness in an HBP1-dependent manner. Together, the emerging mechanism is that EGCG blocks Wnt signaling by inducing the HBP1 transcriptional repressor and inhibits aspects of invasive breast cancer. These studies provide a framework for considering future studies in breast cancer treatment and prevention.


Genetic and biochemical de-regulation of Wnt signaling is correlated with breast and other cancers. Our goal was to identify compounds that block Wnt signaling as a first step toward investigating new strategies for suppression of invasive and other breast cancers.
In a limited phytonutrient screen, EGCG ((؊)-epigallocatechin 3-gallate), the major phytochemical in green tea, emerged as an intriguing candidate. Epidemiological studies have associated green tea consumption with reduced recurrence of invasive and other breast cancers. Wnt signaling was inhibited by EGCG in a dose-dependent manner in breast cancer cells. The apparent mechanism targeted the HBP1 transcriptional repressor, which we had previously characterized as a suppressor of Wnt signaling. EGCG treatment induced HBP1 transcriptional repressor levels through an increase in HBP1 mRNA stability, but not transcriptional initiation. To test functionality, DNA-based short hairpin RNA (shRNA) was used to knockdown the endogenous HBP1 gene. Consistently, the HBP1 knockdown lines had reduced sensitivity to EGCG in the suppression of Wnt signaling and of a target gene (c-MYC). Because our ongoing studies clinically link abrogation of HBP1 with invasive breast cancer, we tested if EGCG also regulated biological functions associated with de-regulated Wnt signaling and with invasive breast cancer. EGCG reduced both breast cancer cell tumorigenic proliferation and invasiveness in an HBP1-dependent manner. Together, the emerging mechanism is that EGCG blocks Wnt signaling by inducing the HBP1 transcriptional repressor and inhibits aspects of invasive breast cancer. These studies provide a framework for considering future studies in breast cancer treatment and prevention.
Breast cancer afflicts 1 in 8 women over her lifetime. Whereas the widespread use of mammography has greatly improved breast cancer survival, the increasingly frequent detection of earlier breast cancers highlights a need for better prognosis and prevention. The most frequently detected early types are now ductal carcinoma in situ and infil-trating ductal carcinoma (also known as invasive breast cancer) (1). The investigation of mechanisms and candidate compounds that may limit increased proliferation and invasiveness in breast cancers could advance new clinical strategies for prognosis and for preventing progression. In this paper, we investigated the green tea compound EGCG 2 in the suppression of Wnt signaling. These studies highlight how knowledge from disparate disciplines can be used to address a possible link to breast cancer mechanisms. In this section, the relevant background into Wnt signaling and the green tea compound EGCG is summarized.
Constitutive Wnt signaling has emerged as a paradigm in breast and other cancers (reviewed in Refs. 2 and 3). Current knowledge indicates that genetic and biochemical de-regulation can result in aberrant Wnt signaling. The essential event in Wnt signaling is the stabilization of ␤-catenin. The resulting accumulation of ␤-catenin increases the pool of nuclear ␤-catenin bound to transcription factor TCF/LEF in complexes that can activate certain genes that ultimately establish the oncogenic phenotype. In numerous cancers, there are reports of loss-offunction mutations in APC and Axin, and gain-of-function mutations in ␤-catenin (reviewed in Ref. 4). De-regulated Wnt signaling by genetic or biochemical means triggers an oncogenic gene expression program that contributes to breast tumorigenesis (3). Wnt gene targets with clinical interest include Cyclin D1 and c-MYC. Finally, recent studies hint that Wnt signaling may have a broader impact through an autocrine mechanism and may provide another explanation of how ␤-catenin levels may accumulate in the absence of genetic mutation of Wnt pathway components (5).
Wnt1 was the first oncogene that was discovered by mammary-specific murine mammary tumor virus retroviral insertion. Transgenic expression of Wnt remains an established model of breast tumorigenesis. As known G 1 regulators Cyclin D1 and c-MYC are established oncogenes with overexpression in breast cancers. The ongoing definition of new gene targets indicate that constitutive Wnt signaling may affect angiogenesis, invasion, and metastasis, which are important processes in cancer (2). Because of an unexpectedly wide role in breast cancer processes, compounds and proteins that limit Wnt signaling in cancer may have an impact for prevention and treatment (6).
We have reported that the HBP1 transcriptional repressor is a suppressor of Wnt signaling (7). HBP1 is a high mobility group box con-taining transcription factor, like the LEF and TCF transcription factors in the Wnt pathway. HBP1 was also characterized as a transcriptional repressor and cell cycle inhibitor in cells and animals (8 -15). HBP1 suppresses Wnt signaling at the level of the transcriptional activation, thereby preventing the expression of genes that would otherwise establish the oncogenic phenotype. In the context of Wnt signaling, HBP1 suppressed the expression of endogenous Cyclin D1 and c-MYC (7). Our recent studies indicate that other relevant breast cancer pathways may be regulated by the HBP1 transcriptional repressor (Ref. 17 and see "Discussion"). Finally, functional loss of HBP1 has been shown to be associated with invasive breast cancer. 3 Strikingly, mutants/variants of HBP1 were observed in 30% of patients with invasive disease. No HBP1 mutants/variants were observed in either controls or patients with ductal carcinoma in situ. Furthermore, HBP1 mutants/variants were uniformly defective in Wnt signaling and suppression of growth and invasiveness. 3 The sum of both the molecular and clinical data suggests that HBP1 is a candidate gene that may have a role in breast cancer progression, and that HBP1 may be a useful target for prevention and treatment of breast cancer.
To identify possible compounds, we surveyed the epidemiological and nutritional literature to identify natural compounds that may block constitutive Wnt signaling. Epidemiological studies have long linked the consumption of certain foods to reduced cancer prevalence. Phytonutrient compounds from green tea, tomato, red wine, and other foods have been identified. The definition of bioactive food components provides an opportunity to test their role in specific signaling pathways that may have cancer preventative impact. These investigations may also provide new informative biomarkers as indicators of treatment efficacy. Whereas we initially screened several candidate phytonutrients, the focus of this paper is the green tea compound EGCG ((Ϫ)-epigallocatechin gallate), which is the principal bioactive component. Several epidemiological studies have highlighted green tea consumption with decreased occurrence of several cancers (reviewed in Ref. 19). Green tea consumption has been associated with the appearance of less aggressive breast cancer and an overall reduced rate of breast cancer recurrence in Japanese women (20 -22). EGCG is the main catechin component in dry green tea (about 30%). Brewed green tea is about 0.1% EGCG solution (w/v), or up to 2 mM. Green tea and EGCG (equivalent of 4 -8 cups/day) have been tested in humans with no appreciable side effects and in United States phase 1 and 2 trials (Refs. 23 and 24; reviewed in Ref. 25). Several animal studies have begun to define EGCG preventative effects. The oral delivery of EGCG reduced tumor progression in animal models of breast cancer (26,27), but no molecular mechanisms were addressed in these studies. Last, EGCG contributes to suppression of intestinal tumors in the Min mice, which arise through an Apc mutation and is an animal model of constitutive Wnt signaling. In somatic colon cancer, the prevalent APC mutations represent the earliest defined stage in the colon cancer pathway (reviewed Ref. 28 and 29). Together, EGCG is a very intriguing compound for cancer prevention.
Our results indicate that EGCG blocked Wnt signaling through the HBP1 transcriptional repressor that was previously shown to inhibit Wnt signaling. The data in this paper supports a model by which EGCG induces HBP1 by stabilizing its mRNA and not through transcriptional induction. EGCG-induced HBP1 blocked Wnt signaling and cellular processes that are relevant to invasive breast cancer. Collectively, these studies further establish a role of HBP1 in suppressing Wnt signaling and define another signaling context for future EGCG applications in breast cancer prevention and treatment.
Plasmids-The pEFBOS vectors were used to express hemagglutinin (HA) epitope-tagged versions of human HBP1 wild type and mutants, which were published (13,16). HA epitope-tagged HBP1 cDNA was subcloned into the pBabe-Puro vector for the construction of HBP1 overexpressing cell lines. TOPFLASH, FOPFLASH, CMV-HA-␤-cate-nin⌬45, and CMV Wnt1 were previously described (see Ref. 7). HBP1 shRNA were provided by the DNA-based shRNA-expressing retroviral vector pSR (Oligoengine). Three shRNA coding oligos against human HBP1 were designed and verified to be specific to HBP1 by Blast search. The most effective HBP1 small interfering RNA target sequence is ACTGTGAGTGCCACTTCTC, which was used in all subsequent experiments. DNA encompassing 2 kb upstream to ϩ83 bp of the HBP1 transcription start site was cloned into the pGL3basic luciferase vector (gift of Michael A. McDevitt).
Viral Production and Infection of Target Cells-Amphotropic retroviral supernatants were produced by transfection of phoenix packaging cells by FuGENE 6 (Roche Diagnostics). After 48 h, the tissue culture medium was filtered through a 0.45-m filter. The viral supernatant was used for infection of cells after addition of 8 g/ml Polybrene (Sigma). Cells were infected for 3 h and allowed to recover for 24 h with fresh medium. The infected cells were selected with puromycin.
SYBR Green Real-time RT-PCR-The HBP1 RT-PCR forward primer (EX 4-5) was ATCATCTCCTGTACACATCATAGC and the reverse primer (EX 7-3) was CATAGAAAGGGTGGTCCAGCTTAC. The HBP1 pre-mRNA RT-PCR forward primer is EX 4-5 as above, whereas the reverse primer was AGTTTAACCTCATAAATAAACTTAC (Intron 4-3). cMyc forward and reverse primers for real-time RT-PCR were 5Ј-GGCTTTATCTAACTCGCTGTAG and 3Ј-GAGTCG-TAGTCGAGGTCATAGTTC. 18 S forward and reverse primers were GTCTGTGATGCCCTTAGATG and AGCTTATGACCCGCACT-TAC. Total RNA was extracted using RNAeasy Mini kits (Qiagen) and reverse transcribed with the random hexamer mixture (iScript, Bio-Rad). The resulting cDNA were used for real-time PCR performed on Opticon (MJ Research). SYBR Green master mixture was purchased from Bio-Rad. All quantitations were normalized to an endogenous 18 S RNA control. The relative quantitative value for each target gene compared with the calibrator for the target was expressed as comparative Ct (2 Ϫ(⌬Ct-Cc) ) method (Ct and Cc are the mean threshold cycle differences after normalizing to 18 S). Real-time RT-PCR experiments were performed in triplicate and the percent coefficient of variation for the set value was less than 0.1. PCR efficiency with given primers was between 95 and 105%. Melting curves were also performed for identification of primer-specific amplicons; the HBP1 mRNA melting temperature T m was 86°C, the HBP1 pre-mRNA (exon 4-inton 4) T m was 84.5°C, whereas the 18 S T m was 90°C.
Reporter Assays-A 100-mm plate was harvested 48 h after transfection and lysed using 300 l of Promega lysis buffer (100 mM sodium phosphate, 0.2% Triton X-100, pH 7.5) supplemented with 200 M phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, and 1 g/ml leupeptin. After three freeze-thaw cycles, lysates were rocked for 30 min at 4°C and then centrifuged for 30 min at 14,000 ϫ g. The amount of luciferase enzyme in each lysate was determined using the Luciferase Assay System (Promega) according the manufacturer's protocol. Transfection efficiency was normalized using transfected RSV-␤-gal (Rous sarcoma virus vector with ␤-galactosidase). ␤-Galactosidase activity was quantitated using the colorimetric substrate o-nitrophenyl-␤-D-galactopyranoside (Sigma).
Western Blotting-Whole cell lysates were prepared from pelleted cells by extraction for 30 min at 4°C in lysis buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM sodium van-adate, 1 g/ml leupeptin, 200 M phenylmethylsulfonyl fluoride, 1 g/l pepstatin). Cytoplasmic lysates were prepared identically except using cytoplasmic lysis buffer (20% glycerol, 20 mM HEPES, pH 7.6, 1.5 mM MgCl 2 , 0.1% Triton X-100, 10 mM NaCl, 0.2 M EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 1 g/ml leupeptin, 10 g/ml aprotonin, 10 g/ml pepstatin A, 10 mM ␤-glycerophosphate, 1 mM sodium vanadate). Nuclear lysates were then prepared from the cytoplasmic pellet using nuclear lysis buffer (20% glycerol, 20 mM HEPES, pH 7.6, 1.5 mM MgCl 2 , 0.1% Triton X-100, 500 mM NaCl, 0.2 M EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 1 g/ml leupeptin, 10 g/ml aprotonin, 10 g/ml pepstatin A, 10 mM ␤-glycerophosphate, 1 mM sodium vanadate). All lysates were clarified by centrifugation for 15 min at 14,000 ϫ g and frozen at Ϫ80°C. For detection of transfected HA-tagged HBP1, the HA.11 antibody (Covance) was used at 1:1000 dilution. The polyclonal rabbit antibody to HBP1 was as previously described (12) or was purchased (Santa Cruz Biotechnology). The ␤-catenin antibody was obtained from Upstate Biotechnology. The phosphoserine 33/41 ␤-catenin was purchased The results are normalized for transfection efficiency and reported as -fold activation. Each bar represents the mean Ϯ S.E. The TOPFLASH reporter contains LEF/TCF sites that are used routinely to detect Wnt signaling activity. The mutated reporter FOPFLASH was also included as a control. EGCG inhibited Wnt1-activated reporter activity in a dose-dependent manner. Values with * or ** above the bars are significantly different from the control untreated cells (p Ͻ 0.05 or p Ͻ 0.01, respectively). B, EGCG suppressed Wnt signaling in non-breast cells (HEK 293T). Wnt signaling was activated on the co-expressed reporter TOPFLASH using either Wnt1 expression vector (left panel) or LiCl inhibition of GSK-3␤ (right panel). The results were normalized for transfection efficiency and expressed as -fold activation. Each bar represents the mean Ϯ S.E. The mutated reporter FOPFLASH was also included as a control. EGCG inhibited both Wnt1-and LiCl-activated reporter activities in a dose-dependent manner. In both the left and right panels, values with * or ** above the bars are significantly different from the control untreated cells (p Ͻ 0.05 or p Ͻ 0.01, respectively). C, EGCG suppressed Wnt signaling in constitutively activated cells. The HCT 116 has highly activated Wnt signaling due to a genetic mutation of ␤-catenin. Thus, EGCG regulation of the TOP-FLASH reporter could be measured without prior treatment and as described in B and C. EGCG also inhibited TOPFLASH reporter activity in a dose-dependent manner. In the figure from Cell Signaling. Western blots were visualized using enhanced chemiluminescence (PerkinElmer Life Sciences).
Statistical Analysis-All statistical analyses were done by two-way analysis of variance with the Tukey test, using SAS program version 9.12.

RESULTS
The Green Tea Polyphenol EGCG Inhibits Wnt Signaling and the Wnt Target Gene c-MYC-We initially screened several bioactive compounds from foods and diets that are linked to reduced cancer risk for the suppression of Wnt signaling. These compounds included EGCG (green tea), curcumin (turmeric), resveratrol (red wine), and lycopene (tomato). Based on initial results, EGCG was selected for further analysis. To delineate the target of action in the canonical Wnt pathway, the impact of EGCG was assessed in cells in which Wnt signaling was activated at different points (7). Wnt signaling was activated by either: 1) expression of Wnt, or 2) through inhibition of GSK-3␤ using LiCl or SB 415286 (30,31). All protocols give increased ␤-catenin stability and robust transcriptional activation through LEF/TCF sites (assessed by the TOPFLASH reporter consisting of three LEF/TCF sites fused to a minimal c-FOS promoter). The FOPFLASH reporter containing mutant LEF/TCF sites was used as a negative control. Finally, we tested the impact of EGCG on expression of the c-MYC gene, a documented Wnt target. These approaches provide a clear and simple framework for identifying any effects of EGCG on Wnt signaling. Whereas the primary focus of this paper is breast, it is important to verify results in other cell types to address generality.
As shown in Fig. 1A, in MDA-MB-231 human breast cancer cells, the TOPFLASH reporter was strongly induced by either transfected Wnt, whereas the FOPFLASH control was only marginally affected by either activator. Data bars represent -fold induction of the reporter by the stimulus relative to the basal reporter activity. EGCG treatment of MDA-MB-231 cells suppressed Wnt signaling in a dose-dependent manner from 25 to 100 M, whereas there was no change in the control FOPFLASH reporter. Importantly, similar results were obtained, regardless of whether the Wnt pathway was activated by transfected Wnt or transfected ␤-catenin. Similar results for EGCG inhibition were observed in 293T cells activated by Wnt or GSK-3␤ inhibition (Fig. 1B) and in NIH 3T3 cells (not shown). Last, the results with HCT116 cells show that constitutive Wnt signaling (due to a stabilizing ␤-catenin mutation) is also repressed by EGCG (Fig. 1C).
To corroborate the signaling results, we next investigated the c-MYC gene as a relevant Wnt target gene (32) in which increased expression is linked to breast and other cancers. c-MYC has a defined role in G 1 control and is an established oncogene. Recent studies have indicated that c-MYC has a key role in human cell transformation in oncogene complementation studies and in the regulation of the telomerase gene, which is a critical component in transformation (33,34) (reviewed in Ref. 35). Previously, we showed that the expression of endogenous c-MYC was decreased by HBP1 overexpression and upon suppression of Wnt signaling (7). As measured by quantitative real-time RT-PCR (Fig. 1D), activation of Wnt signaling by inhibition of GSK-3␤ induces c-Myc expression. Like the Wnt signaling experiments in Fig. 1, A-C, the induction of the c-MYC target gene was also blocked by EGCG. The conclusion of Fig. 1 for both breast and non-breast lines is that EGCG can block Wnt signaling. Whereas the focus of most of the later experiments in the paper will be breast cancer, the results may have a wider impact on other cancers.

HBP1 Contributes to the Suppression of Wnt Signaling and Gene
Expression by EGCG-The experiments of Fig. 1 suggested two testable mechanisms for EGCG blocking Wnt signaling: 1) EGCG might trigger a reduction in ␤-catenin levels, or 2) EGCG might increase the levels of HBP1. Previously, we showed that HBP1 is a suppressor of Wnt signaling by inhibiting the LEF/TCF/␤-catenin transcriptional activation complex (7). As shown in Fig. 2A, endogenous ␤-catenin levels were unchanged with EGCG treatment. In addition, both the phosphorylation of ␤-catenin by GSK-3␤, as well as total, cytoplasmic and nuclear ␤-catenin levels were unchanged (Fig. 2B), further supporting the lack of change in ␤-catenin levels with EGCG. However, we observed that in 293T cells ( Fig. 2A) and in MDA-MB-231 cells (Fig. 4A), endogenous HBP1 protein levels increased with EGCG treatment. These results were extended with the observation that EGCG treatment increased HBP1 mRNA, as measured by real-time RT-PCR (Fig. 3A). Thus, EGCG increased expression of HBP1, suggesting that HBP1 may be a plausible target of EGCG action in the inhibition of Wnt signaling.
We investigated the mechanism by which EGCG induces HBP1 mRNA and protein. We considered the possibilities that EGCG increased 1) HBP1 gene transcription, or 2) HBP1 mRNA stability. Endogenous HBP1 mRNA was induced 4-fold by EGCG (Fig. 3A). Thus, the next test of EGCG regulation was an analysis of HBP1 promoter activity. EGCG did not appear to activate a 2-kb HBP1 promoter reporter construct (Fig. 3B). In addition, we measured the levels of nascent HBP1 pre-mRNA in the presence and absence of EGCG. A quantitative PCR adaptation of nuclear run-on assays has been recently reported (18) and was used to measure unspliced, HBP1 pre-mRNA. The key step is designing primers that detect exon-intron junctions that are specific to pre-mRNAs. If the effect were transcriptional, then the levels of unspliced or pre-HBP1 mRNA would be altered, in addition to the mature HBP1 mRNA. We measured unspliced HBP1 pre-mRNA levels in control and EGCG-treated cells. Several exon and adjacent intron primer sets were designed for quantitative real-time RT-PCR. Fig. 3C show representative and quantified data measuring pre-mRNA from exon 4 into intron 4. These results demonstrate no change in HBP1 pre-mRNA levels with EGCG treatment, whereas mature HBP1 mRNA increased. Finally, we analyzed EGCG regulation of HBP1 mRNA stability. Actinomycin D was used to block new RNA synthesis, so that decay of existing transcripts could be detected. As shown in Fig.  3D, the half-life of HBP1 mRNA in untreated cells was 32 min. However, EGCG treatment increased HBP1 mRNA half-life, such that it was sta-FIGURE 3. HBP1 mRNA is stabilized by EGCG treatment. A, quantified HBP1 mRNA levels using real-time RT-PCR. Four independent real-time RT-PCR analyses were performed as described under "Materials and Methods." HBP1 mRNA was normalized to 18 S RNA and the relative HBP1 mRNA levels were plotted on the graph. 50 M EGCG gave a significant induction of HBP1 mRNA expression as indicated (*, p Ͻ 0.01). B, HBP1 promoter analysis. A Ϫ2-kbp HBP1 promoter/luciferase reporter construct (see "Materials and Methods") was transfected in 293T cells and treated with 25 or 50 M EGCG. The results are normalized for transfection efficiency and reported as -fold activation. Each bar represents the mean Ϯ S.E. The empty pGL3 Basic reporter was also included as a negative vector control. EGCG had no effect on either the HBP1 reporter or the empty vector. C, real-time RT-PCR analysis of EGCG regulation of HBP1 pre-mRNA. In experiments similar to Fig. 3A, HBP1 pre-mRNA and mature RNA (see "Materials and Methods") from 50 and 100 M EGCG-treated and untreated 293T cells were quantified using real-time RT-PCR. Only mature HBP1 mRNA was induced as indicated by the asterisk (*, 50 M EGCG, p Ͻ 0.05, **,100 M EGCG, p Ͻ 0.01). D, EGCG regulation of HBP1 mRNA half-life. Control and 50 M EGCG-treated 293T cells were incubated with actinomycin D for the times indicated, then total RNA from the cells was isolated and quantified by real-time RT-PCR. The half-life (t1 ⁄2 ) of HBP1 mRNA was determined to be 32 min. The presence of 50 M EGCG stabilized HBP1 mRNA such that the half-life could not be determined in this short-term experiment. Both the 30-and 60-min time points were significantly different from controls (*, p Ͻ 0.05). APRIL 21, 2006 • VOLUME 281 • NUMBER 16 ble over the time course of the experiment (1 h). Taken together, these results demonstrated that EGCG increased the stability of HBP1 mRNA, resulting in the observed increase in HBP1 protein. Similar results were obtained in several cell lines (293T, MDA-MB-231, and NIH 3T3 (Fig. 4A)), so the stabilization of HBP1 appears to be a general response to EGCG. However, the effects of EGCG on increasing HBP1 mRNA stability do not represent a general RNA stabilization, as the c-MYC RNA is decreased upon EGCG treatment (see Figs. 1D and 5C). Thus, a plausible hypothesis is that EGCG may increase HBP1 mRNA as part of the mechanism for inhibition of Wnt signaling.

HBP1, Green Tea, EGCG, and Suppression of Wnt Signaling
Whereas EGCG appeared to induce HBP1 mRNA and protein levels, we next tested whether HBP1 functionally contributes to the observed suppression of Wnt signaling by EGCG. The strategy was to knockdown HBP1 and to assess the consequences on EGCG-mediated regulation of Wnt signaling. We used DNA-based RNA interference to knockdown the HBP1 mRNA and protein levels. The pSuper-retro vector (Oligoengine) was used to create stable lines expressing HBP1 shRNA, which is subsequently processed to small interfering RNAs. As shown in Fig. 4, A and B, this approach reduced the overall expression of HBP1 protein and mRNA in NIH 3T3 and MDA-MB-231 cells. Expression of the Values with an asterisk (*) above the bars are significantly different from the control untreated cells (p Ͻ 0.05). HBP1 mRNA was induced by EGCG in pSR-control cells, but was strongly decreased in HBP1-shRNA-expressing cells, even with EGCG treatment. C, Wnt signaling is increased with HBP1 knockdown. Wnt signaling was measured in pSRcontrol and HBP1-shRNA cell lines (NIH 3T3 and MDA-MB-231) by expression of Wnt1 and TOP-FLASH activity as described in the legend to Fig. 1. Each bar represents the mean Ϯ S.E. Wnt-activated gene expression in HBP1-shRNA cells was considerably higher relative to controls. These results are consistent with the notion that HBP1 is a suppressor of Wnt signaling. Values with an asterisk (*) above the bars are significantly different from the control untreated cells (p Ͻ 0.05). IB, immunoblot.

HBP1, Green Tea, EGCG, and Suppression of Wnt Signaling
HBP1-shRNA also significantly reduced the ability of EGCG to induce both HBP1 mRNA and protein levels (Fig. 4, A and B). The control shRNA did not change HBP1 expression. We initially characterized the HBP1 knockdown lines with respect to Wnt signaling using a reporter assay (TOPFLASH). Consistent with our original observation that HBP1 is a suppressor of Wnt signaling (7), the knockdown of HBP1 conferred increased sensitivity to Wnt signaling in NIH 3T3 and MDA-MB-231 cells (Fig. 4C). Thus, the HBP1 knockdown lines exhibited the expected functional consequences with respect to Wnt signaling and provided a good setting to test EGCG and possibly other inhibitory compounds for future studies.
We next examined the consequences of reducing HBP1 on the suppression of Wnt signaling by EGCG. Using submaximal EGCG concentrations, the HBP1 knockdown cell lines had reduced sensitivity to EGCG for suppression of Wnt signaling. As shown in Fig. 5, A and B, EGCG suppressed transfected Wnt-induced signaling in the control, but not in independently isolated HBP1 knockdown lines in both MDA-MB-231 breast and NIH 3T3 cells. These results supported the generality of the EGCG response and the dependence on HBP1. It should be noted that EGCG treatment and RNA interference knockdown are opposite processes with respect to regulating HBP1 levels (Fig. 4, A and  B). However, apparent HBP1 levels in the knockdown lines are still less than control cells even with EGCG stimulation. Thus, these lesser HBP1 levels in RNA interference-expressing cells likely contribute to the reduced inhibition of Wnt signaling by EGCG in the knockdown cells. As shown in Fig. 5, A and B, there is a difference in the dose required for 50% suppression in the MDA-MB-231 (50 M) and NIH 3T3 (25 M) cells and likely reflects a differential sensitivity to EGCG. Finally, there was no significant inhibition of c-MYC mRNA by EGCG treatment in the HBP1 knockdown line (Fig. 5C). Together, these data support the notion that EGCG sensitivity depended upon HBP1 levels.
EGCG Regulates Features of Invasive Breast Cancer-Epidemiological studies have suggested that green tea consumption is associated with reduced recurrence in patients with invasive breast cancer. We have shown that HBP1 function is abrogated in patients with invasive breast cancer, and that HBP1 regulates biological process associated with invasive breast cancer, including proliferation (17, 41) and invasion. 3 Furthermore, aberrant Wnt signaling is also associated with proliferation and invasion (42,43).
We tested the effect of EGCG on both proliferation and invasiveness in MDA-MB-231 breast cancer cells, which is a model for invasive breast cancer. In addition, MDA-MB-231 cells have constitutive and autocrine Wnt signaling, which likely contributes to the high tumorigenic growth and invasiveness. Thus, these cells are ideal for investigating the role of HBP1 and EGCG in suppressing biological properties associated with invasive breast cancer. As shown in Fig. 6A, EGCG treatment resulted in decreased migration of MDA-MB-231 cells toward fibronectin-treated media. Using 50 and 100 M EGCG concentrations, there was, respectively, an ϳ50 and 70% reduction of migration. As shown in Fig. 6B, EGCG treatment resulted in decreased invasion of MDA-MB-231 cells through matrigel. Intriguingly, the HBP1 knockdown line showed increased migration and invasiveness relative to the control line. EGCG still suppressed both migration and invasion in the presence of reduced HBP1 levels, but less effectively. The concentration profiles for EGCG are grossly similar to Figs. 4 and 5, although somewhat lower EGCG was required for inhibition of invasion and migration. Thus, EGCG can suppress migration and invasion of breast cancer cells, but with reduced efficacy upon HBP1 knockdown.
EGCG also suppressed breast cell proliferation and transformation, as measured by anchorage independent growth in soft agar. MDA-MB-231 cells are an established model of breast cell transformation.
Although not always the case in human cells, growth in soft agar does correlate with tumorigenic growth for MDA-MB-231 cells. Previous studies have shown that green tea can inhibit MDA-MB-231 tumor formation in xenograft models at achievable consumption levels of ϳ5 cups/day, but the mechanisms were not reported (26). In these studies, it was not clear whether tumor establishment, proliferation, or both were abrogated. As shown in Fig. 7, A and B, tumorigenic proliferation was clearly affected. Cells with HBP1 knockdown resulted in larger and more numerous colonies, in comparison to the control MDA-MB-231 cells. EGCG inhibited soft agar colony formation of both control and HBP1 knockdown lines, but to significantly different degrees. With 30 M EGCG, there was a 6-fold reduction in colony formation, whereas the colony numbers in the HBP1 knockdown line were reduced by less than 2-fold. This result suggested that HBP1 contributed to EGCGmediated inhibition of tumorigenic growth in MDA-MB-231 cells. Thus, at the minimum, EGCG affects tumorigenic proliferation. Whether EGCG also affects the establishment of the xenograft tumors remains to be investigated. Together, Figs. 6 and 7 indicate that HBP1mediated EGCG function can attenuate two important features of invasive breast cancer and begins to provide a molecular basis for the epidemiological observations.

DISCUSSION
Numerous observations have indicated that constitutive Wnt signaling is a factor in breast and other cancers. The objective of this study was to bridge divergent disciplines to identify possible new breast cancer prevention strategies. We sought to identify natural compounds that may be used for blocking cancers with constitutive Wnt signaling. As described, the initial screen showed that EGCG was the best of five natural compounds for suppressing Wnt signaling in the cell-based assay. Epidemiological and some cellular studies indicate that green tea and EGCG could have preventative properties in breast and other cancers (19,36). The mechanism of EGCG-mediated suppression of the Wnt pathway was systematically investigated (Fig. 8). A previous paper (37) has suggested that EGCG may block ␤-catenin accumulation, but we found no evidence for this mechanism. A possible resolution is that the previous study examined exogenous transfected ␤-catenin, whereas our study examined levels of endogenous ␤-catenin. We found no change in the levels of total ␤-catenin, and no changes in the ␤-catenin population that is phosphorylated at the GSK-3␤ site (Fig. 2, A and B).
Our work demonstrating that HBP1 inhibits Wnt signaling (7) suggested another mechanism for EGCG regulation of Wnt signaling (summarized in Fig. 8). As shown in Figs. 1-3, a systematic evaluation revealed that HBP1 is a probable target of EGCG action, which results in the stabilization of HBP1 mRNA and an increase in HBP1 protein.
These results are significant because our previous work linking HBP1 to suppression of Wnt signaling (7) suggests that EGCG-mediated induction of HBP1 is a plausible mechanism for suppression of this pathway and at least one critical target gene (c-MYC, Figs. 1D and 5C). These observations, together with the experiments demonstrating that an HBP1 knockdown has diminished sensitivity to EGCG (Fig. 5, A-C), suggest that HBP1 may be an attractive endogenous target for nutritional intervention.
Several recent studies highlight the potentially wide impact of Wnt signaling in tumorigenesis and for the development of new strategies to limit constitutive Wnt signaling. A new strategy necessarily requires a biochemical and mechanistic understanding of the action of EGCG and other candidate compounds and for the improved design of more advanced preclinical studies. Constitutive Wnt signaling has been linked to increased proliferation and invasiveness in breast and other cancers. In breast cancer, excessive ␤-catenin levels are correlated to breast cancers of poor prognosis (38 -40), whereas the overall percentage of tumors with high ␤-catenin expression is estimated to be 50% (3). Unlike many other cancers, no mutations in APC, Axin, or ␤-catenin have been reported for breast cancers. In fact, many cancer types have high ␤-catenin levels with an apparently genetically intact Wnt pathway. In a possible clue, a recent paper (41) has reported that many cancer cells secrete Wnt family ligands, which then activate the Wnt pathway through the frizzled receptors. The consequence is a genetically intact, but constitutive pathway. Compelling evidence for this mechanism demonstrated that frizzled inhibitors blocked high ␤-catenin levels (41).
Wnt signaling also appears to play a role in other aspects of cancer pathophysiology, including invasion, metastasis, and resistance to treatment. For example, expression of Wnt family ligands has been clinically correlated with invasive breast cancer (3). In recent studies, Twist has been shown to play a significant role in metastasis in breast cancer (16). Twist expression is regulated by Wnt/␤-catenin (49). Last, Wnt signaling has been linked to stem cell renewal in breast and other tissues. Mouse mammary models of Wnt signaling have highlighted that mouse mammary cells with constitutive Wnt signaling possess stem cell characteristics. There is considerable speculation that breast cells with constitutive Wnt signaling may lead to a treatment-refractory tumor population (reviewed in Refs. 4 and 6).
The importance of Wnt signaling in breast cancer is further highlighted by our recent observation that HBP1 is abrogated in invasive breast cancer. 3 In a screen of 76 archived breast cancer specimens, we identified 10 HBP1 mutations/variants that were associated with fully invasive breast cancer. Furthermore, all of the HBP1 mutants/variants were defective in inhibition of Wnt signaling. These findings suggest a new mechanism for generating increased Wnt signaling in breast tumors through decreased inhibition. Finally, the fact that HBP1 abrogation occurred only in invasive disease provides an additional mechanism linking Wnt signaling with invasive breast cancer.
In contemplating future clinical applications, the identifications of biomarkers for EGCG action would aid in study design. Whereas we have focused on the mechanisms, this study may implicate the reduction of c-MYC mRNA and the induction of HBP1 mRNA as two possible criteria for EGCG-mediated suppression of Wnt signaling. Whereas c-MYC is a biologically relevant target, it should be noted that not all Wnt targets respond to EGCG suppression (not shown). The basis is not clear, but most Wnt target genes have complex promoters (e.g. Cyclin D1) and may respond to many signals within a cell. Wnt signaling and c-MYC expression have been increasingly associated with invasive breast cancers. Elevated c-MYC levels (in either the presence or absence of significant gene amplification) were found in infiltrating ductal carcinomas, rather than normal or ductal carcinoma in situ (42)(43)(44). Thus, c-MYC may be a very useful biomarker in the breast cancer context.
An important limitation of our interpretations is that both EGCG and HBP1 may affect other pathways relevant to breast cancer and that may also intersect the Wnt pathway. Specifically, there are reports that EGCG may FIGURE 8. Schematic of Wnt signaling regulation by EGCG. The Wnt signaling pathway regulates the relative stability of ␤-catenin via GSK-3␤dependent phosphorylation. Wnt signaling leads to the inhibition of GSK-3␤ and decreased phosphorylation of ␤-catenin, leading to increased stability. The unphosphorylated ␤-catenin is translocated into the nucleus and combines with the human menopausal gonadotropin box transcription factors LEF and TCF to activate target genes such as c-MYC. Our work has shown that the HBP1 transcriptional repressor is a suppressor of Wnt signaling. Many of the regulatory proteins can be oncogenes or tumor-suppressor genes (designated by an asterisk). In the experiments outlined, Wnt signaling in cells is stimulated by transfected Wnt or through GSK-3␤ inhibition. Each of these activation points is indicated in a box. In the present study, EGCG may down-regulate Wnt signaling target gene expression by activating the Wnt pathway negative regulator HBP1 resulting in reducing invasive breast cancer. diminish signaling through the epidermal growth factor receptor family, which has a critical role in breast cancers with poor prognosis and to invasiveness. The specific effect of EGCG is a decrease in EGFR tyrosine phosphorylation and diminution of downstream signaling. In the context of defining other target genes for HBP1, we have identified p47 phox as a target gene that may also indirectly modulate EGFR phosphorylation (17). Last, several papers have recently linked Wnt and EGFR signaling pathways in different ways (45,46). Thus, the intersection of the EGFR and Wnt signaling pathways, which are both important in breast cancer, requires further study in the context of EGCG and the HBP1 suppressor (45). Remarkably, HBP1 regulates G 1 progression and EGFR signaling, both of which have been attributed to EGCG. Our current studies are aimed at a direct comparison of EGCG and HBP1 in relevant biological processes to test if HBP1 might be the major target for EGCG. In the current study, the link of EGCG to Wnt signaling provides yet another similarity.
Our study with EGCG opens up future work for combining EGCG with other small molecules. Some small molecule inhibitors of Wnt signaling have been reported with differing mechanisms. These include non-steroidal anti-inflammatory drugs, Cox-2 selective inhibitors, and new proprietary compounds from a chemical screen (47,48). Because our study has characterized EGCG in the context of Wnt signaling, a future question is additivity and/or synergism with other possible compounds. We have shown previously that HBP1 blocks Wnt signaling through blocking transcriptional activation, rather than the regulation of ␤-catenin levels. A future possible combination study may be Cox-2 inhibitors and EGCG and with reduced levels of both compounds to limit possible cardiovascular side effects in the Cox-2 inhibitors. Celebrex and other Cox-2 inhibitors were used to prevent colon cancer in familial adenematous polyposis patients, which have a high risk of colon cancer development because of constitutive Wnt signaling.
The increasing detection of invasive breast cancer highlights a major need for future design of safer prevention and suppression strategies. Our results have implications for suppressing breast cancer proliferation and invasiveness, which are critical issues in invasive breast cancer. Our results further support the role of HBP1 in suppressing Wnt signaling and breast cancer and advance EGCG for specific applications in blocking Wnt signaling. The work in this paper indicates a possible new mechanism of EGCG suppression of breast cancer through the induction of the HBP1 protein and suppression of Wnt signaling. In fact, both c-MYC and HBP1 may be useful biomarkers for future work with EGCG in the context of Wnt, and possibly other signaling pathways. Our study provides a framework for testing other nutrient and pharmacological compounds for blocking Wnt signaling and for potential in cancer prevention or therapeutics.