Repression of WT1-Mediated LEF1 Transcription by Mangiferin Governs β-Catenin-Independent Wnt Signalling Inactivation in Hepatocellular Carcinoma Open Access

Liver malignancy is the second leading cause of cancer mortality, after pulmonary carcinoma, and accounted for an estimated 8.1 million deaths worldwide in 2015 [1]. Hepatocellular carcinoma (HCC) is the most common liver malignancy, accounting for 85% of all liver malignancies, followed by cholangiocarcinoma (∼10%), angiosarcoma (< 1%), and hepatoblastoma. HCC is also categorized as a global disease because it may develop in patients with hepatitis B or C virus infections or with other non-viral risk factors such as obesity, diabetes, and alcohol-induced liver disease [2]. The current optimal therapeutic strategy for HCC is surgical resection and liver transplantation. However, these strategies are applicable in only early stage of HCC with relatively small tumour nodules and good liver function as classified by Child-Pugh score [3]. The asymptomatic and rapid growth of HCC has generally led to poor prognosis and low overall survival rates in patients. The short tumour doubling time and rapid tumour progression in advanced stages of HCC remains the major reason for death [4]. Therefore, there is a great need for anti-tumour agents that decelerate HCC tumour progression and work as a pre-transplantation intervention.

Hyper-activation of the Wnt signalling pathway in HCC has been implicated in tumour progression and aggressiveness [5-7]. Yamashita et al. also postulated that the Wnt pathway is indispensable for maintaining the activities of tumour initiating cells [8], and their Wnt-stimulated gene expression pattern was associated with reduced overall survival in post-surgical HCC patients [9]. The Wnt signalling pathway is initiated by the binding of Wnt ligand to the FZ-LRP5/6 receptor that induces a cascade of events that facilitates β-catenin accumulation and stabilization [10]. The stabilized β-catenin translocates into the nucleus and interacts with the transcription factor TCF/LEF1 for subsequent Wnt target gene activation [11]. As the hallmark of Wnt signalling, β-catenin mutations have been detected in more than 40% of HCC patients [12] and are highly expressed in HBV-related HCC samples [13]. In addition, cytoplasmic retention of β-catenin is associated with poorer prognosis in HCC [14]. Although critical reviews have indicated that β-catenin plays a role in carcinogenesis [15, 16], β-catenin-overexpressing mice showed no hepatic tumour formation [17, 18]. In addition, several instances have reported the activation of β-catenin-independent Wnt signalling cascade, in which TCF/LEF1 may interact with other proteins such as ALY and Smad3 for subsequent gene transduction [19]. Although little is known about TCF/LEF1 in HCC, it has been shown to be a pre-requisite in Wnt signalling and its function is closely associated with HCC malignancy [19].

Mangiferin is a glucosylxanthone abundantly available in the leaves and bark of Mangifera indica L., a species in the family of Anacardiaceae, commonly known as mango [20]. Previous pharmacological studies have reported an inhibitory effect of mangiferin on different cancer cell lines such as those from breast cancer, leukaemia, astroglioma, and lung cancer [21]. The anti-tumour effect of mangiferin was primarily attributed to its suppressive effect on inflammation, oxidative stress, and cell-cycle related proteins [22]. Notably, results from a recent study suggested that mangiferin suppressed the active form of β-catenin, an essential element in the activation of the Wnt pathway [23]. Nonetheless, how Wnt signalling is regulated by mangiferin in inhibiting tumour growth remains unknown.

In this study, we explored the in vivo anti-tumour effect of mangiferin using an orthotopic HCC mouse model with luciferase-expressing cells. We also systemically examined the in vitro inhibitory effect of mangiferin on the survival, growth, migration, and invasion of HCC cells. Molecular target identification was further performed by PCR based-array and Gene Ontology analysis. The transcriptional regulation of a direct target of mangiferin, lymphoid enhancer binding factor 1 (LEF1) by Wilms’ tumour 1 protein (WT1), in the presence and absence of mangiferin, was also investigated.

The human HCC cell line MHCC97L, which was tagged with a luciferase reporter gene, was gifted by Professor Man Kwan from the Department of Surgery, The University of Hong Kong. The HLF cell line was purchased from Japanese Collection of Research Bioresources cell bank. Both cell lines were cultured in Dulbecco’s modified Eagle’s medium (Gibco, Waltham, MA, USA) supplemented with 4.5g/L glucose, 10% foetal bovine serum (Gibco, MA, USA) and 1% penicillin/streptomycin (Gibco, MA, USA)); and they were kept in a 37°C humidified incubator (Thermo Fisher Scientific Inc., Waltham, MA, USA) under 5% CO₂. The DNA plasmid encoding pBABE-puro LEF1 was gifted by Joan Massague (plasmid #27023; Addgene Inc., Cambridge, MA); and the DNA plasmid expressing pAd/WT1-IRES-nAmCyan was gifted by Edward McCabe (plasmid #29756, Addgene). Mangiferin was purchased from Sigma-Aldrich Co. (St. Louis, MO, USA). The stock solution of mangiferin (100 mg/mL) was prepared in DMSO (Sigma-Aldrich Co., St. Louis, M.O., USA) and dH₂O (1: 9, v/v). For animal administration, mangiferin (50 mg/kg) was prepared in 0.4% of sodium carboxyl methyl cellulose (Sigma-Aldrich Co., St. Louis, M.O., USA).

Human HCC was implanted in mice as described in our previous study [24]. Briefly, an ectopic xenograft of HCC was first established by subcutaneously inoculating luciferase-tagged MHCC97L cells (5 × 10⁶ cells) into the right flank of an athymic BALB/c-nu/nu mouse. The tumours were allowed to grow to 10 mm in diameter and then the mouse was sacrificed. The subcutaneously grown HCC tumour was removed and cut into 1-mm³ tumour fragments. Bilateral subcostal skin and muscle incisions were made in another BALB/c-nu/nu mice to expose the entire liver. A tumour fragments was then orthotopically implanted into the left lobe of the liver. One week after this surgery, the recipient mouse was then examined using the IVIS Spectrum imaging system (Perkin Elmer Inc., Waltham, MA, USA).) to observe for bioluminescence, taken as an indication of tumour growth. In brief, mice were anaesthetized and intraperitoneally injected with 15 mg/mL VivoGlo luciferin (#P1041; Promega, Madison, WI USA). Mice showing luciferase signals were randomized to receive a control vehicle (n = 5) or mangiferin (50 mg/kg/2 days, orally; n = 5). Tumour growth was monitored weekly with the IVIS Spectrum imaging system. All mice were sacrificed after 5 weeks of receiving the control vehicle of mangiferin and the hepatic tumours were excised for further histological analysis. The animal procedures described in this study were approved by the Committee on the Use of Live Animals in Teaching and Research, The University of Hong Kong, Hong Kong.

BrdU incorporation assay (Thermo Fisher Scientific Inc., Waltham, MA, USA) was performed according to the manufacturer’s instructions. In brief, the cells were incubated with BrdU labelling solution for 2 hours before fixation with 3.7% formaldehyde (Sigma-Aldrich, St. Louis, M.O., USA). BrdU-labelled cells were then permeabilised, washed, and incubated with anti-BrdU antibody overnight. The cells were further incubated with FITC-conjugated secondary antibody and then counterstained with DAPI (Thermo Fisher Scientific Inc., USA). The cells were visualized under a fluorescence microscope (Olympus, Tokyo, Japan) and the percentage of cells with incorporated BrdU was calculated by dividing the number of stained cell by stained plus unstained cells in five areas of each dish.

The vehicle- and mangiferin-treated cells were fixed in cold 70% ethanol. The fixed cells were incubated with 5 µg/mL propidium iodide (Thermo Fisher Scientific Inc., Waltham, MA, USA) for 30 min prior to analysis on a FACS Canto II flow cytometer (BD and Co., Franklin Lakes, NJ, USA). Cell cycle distribution was analysed with FlowJo analysis software and the percentage of cells in each cycle was expressed as mean ± standard deviation.

HCC cells (5 × 10² cells) were incubated with vehicle or mangiferin in a humidified incubator for 14 days. At the end of this time, the culture medium in each well was removed, and 0.5% crystal violet (BD, Franklin Lakes, NJ, USA) was added. After 30 min of incubation, the plates were rinsed with tap water and dried in room air. The images of colony-filled wells were captured, and the colonies in each group were counted and analysed.

Total RNA from treated cells was purified using the RNeasy Plus Mini reagent kit (Qiagen, Hilden, Germany) and reverse-transcribed using PrimeScript RT master mix (Takara Co., Shiga, Japan). Then, quantitative PCR was performed using SYBR Green master mix (Takara Co., Shiga, Japan) ) and the RT² Profiler human liver cancer PCR array (Qiagen, Hilden, Germany) on a Light Cycler 480 PCR System (Roche Ltd., Basel, Switzerland), as per the manufacturers’ instructions. The relative expression of a target gene transcript after normalization to a reference GAPDH transcript were calculated. The primer sequences used in the study are shown in Table 1.

Paraffin-embedded sections (thickness, 4 μm) were mounted on slides, deparaffinised with xylene, and rehydrated with decreasing concentrations of alcohol (100% to 70%). For haematoxylin and eosin staining, the slides were incubated in Mayer’s haematoxylin (Sigma-Aldrich Co., St. Louis, M.O., USA) for 5 min, followed by 30 s of incubation in 1% acid alcohol and 0.25% eosin Y solution (Sigma-Aldrich Co., St. Louis, M.O., USA). After staining, slides were mounted on Canada balsam (Sigma-Aldrich Co., St. Louis, M.O., USA) prior to analysis on a BX43 light microscope (Olympus, Tokyo, Japan). The histologic growth pattern of the hepatic tumour was documented.

For immunofluorescence staining, the deparaffinised slides were incubated in pre-heated 10 mM citrate buffer (Sigma-Aldrich Co., St. Louis, M.O., USA) for epitope retrieval prior to 30 min of incubation with 10% goat serum containing blocking buffer. Slides were then incubated overnight at 4°C with anti-CD31 (1: 50, #ab28364; Abcam Inc., Cambridge, United Kingdom) or anti-Ki67 (1: 100, #ab15580; Abcam) antibody. The next day, the tissue was incubated for 2 h with Alexa-Fluor 488 conjugated anti-mouse IgG (1: 500; Invitrogen Corp., Carlsbad, CA USA). The slides were counterstained with DAPI then mounted with fluorescent mounting medium (Dako A/S Glostrup, Denmark). The micro vessel density or Ki67-positive cell count was determined by first screening for the hot spot area under magnification of 40x followed by cell counting at five hotspot areas under a magnification of 200×. The mean count of five microscopic field areas was used as the absolute value of micro vessel density or Ki67-positive cells per high power field.

HCC cells were seeded on confocal dishes (Thermo Fisher Scientific Inc., Waltham, MA, USA) and incubated with vehicle or mangiferin for 48 h. The treated cells were then fixed, blocked, and incubated with anti-LEF1 (C12A5) (1: 200, #2230; Cell Signaling Technology Co., Danvers, MA USA) overnight. The following day, Alexa-Fluor 488 conjugated anti-rabbit IgG (1: 500, Invitrogen, Carlsbad, CA, USA) was applied to the dishes for 2 h prior to counterstaining with DAPI. Specimens were visualized under an LSM 780 confocal microscope (Carl Zeiss AG, Oberkochen, Germany).

HCC cells were seeded on reduced-serum supplemented media before transfection. After one day’s growth, the cells were incubated with 5 μg/μL of plasmid in Lipofectamine 2000 Transfection reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions.

Cell lysates were prepared by lysing HCC cells in RIPA lysis and extraction buffer (Thermo Fisher Scientific Inc., Waltham, MA USA) complemented with proteinase inhibitor cocktail (Sigma-Aldrich, St. Louis, M.O., USA) and phosphatase inhibitor (Cell Signaling Technology, Danvers, MA, , USA). Nuclear extraction of HCC cells was performed using cell extraction buffer (Thermo Fisher Scientific Inc., Waltham, MA, USA) according to the manufacturer’s instructions. Equal amounts of denatured protein or nuclear lysates were subject to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE; Bio-Rad, Hercules, CA) and then electro-transferred onto PVDF membrane (Bio-Rad Laboratories Inc., Hercules, CA). Blocking buffer containing 5% BSA was applied to the membrane for 2 h prior to overnight incubation with the relevant antibodies: anti-LEF1 (C12A5) (1: 1000, Cell Signaling Technology, Danvers, MA, USA), anti-WT1 (C-19) (1: 500, Santa Cruz Biotechnology, Dallas, TX, USA), anti-Lamin B1 (1: 500, ab16048; Abcam, Cambridge, UK) and anti-GAPDH (14C10) (1: 1000, Cell Signaling Technology, USA). HRP-conjugated secondary antibody (1: 2000, Cell Signaling Technology, USA) was applied to the membrane for 2 h and then bands were visualized using the Amersham ECL chemiluminescence detection reagent (Sigma-Aldrich Co., St. Louis, M.O., USA).

ChIP assay was performed using the EZ-Magna ChIP A-Chromatin Immunoprecipitation kit (Millipore, Darmstadt, Germany) according to the manufacturer’s instructions. Briefly, the cells were first cross-linked in formaldehyde and lysed in cell and nuclear lysis buffer. The cell lysates were sonicated for 15 cycles of 7-s and 10-s pulses (40% amplitude) to shear the cross-linked DNA, and then incubated overnight with anti-WT1 (C-19) (Santa Cruz Biotechnology, Dallas, TX, USA) or IgG control, and protein A magnetic beads. WT1/DNA complexes were precipitated, and reverse cross-linked by incubation with elution buffer supplemented with proteinase K at 62°C. The DNA was then purified by applying the sample to a DNA separation column. The binding capacity of WT1 to the LEF1 promoter was analysed by qPCR and the shear DNA sample served as an input control. The promoter binding region was verified using an Applied Biosystems 3730xl DNA analyser (Thermo Fisher Scientific Inc., Waltham, MA, USA).

The microarray dataset under accession number GDS4887 [25] was downloaded from the publicly accessible Gene Expression Omnibus (GEO) database. Expression patterns of LEF1 in tumour and non-tumor regions of HCC specimens were analysed and compared. Gene expression arrays of LEF1 and WT1 in patients with HCC were extracted from the open-access The Cancer Genome Atlas (TCGA) database using cBioPortal [26, 27]. In the cBioPortal interface, the study of “Liver Hepatocellular Carcinoma (TCGA, Provisional)” and genome profiles of “mRNA Expression z-Scores (RNA Seq V2 RSEM)” were selected. For the gene set of interest, “LEF1” or “WT1” was entered. The dataset included data from 442 patients’ samples and the mRNA expression (in log 2 level) was analysed according to the patients’ overall survival and histological tumour grade. The correlation between gene expressions and HCC patients’ overall survival was used to plot Kaplan-Meier curves.

Data from all experiments were expressed as mean ± standard deviation. The data between groups was analysed using a one-way analysis of variance (ANOVA) or student’s t test. A p value less than 0.05 was considered an indicator of statistical significance.

To investigate whether mangiferin suppresses hepatocellular carcinoma growth and progression, the orthotopic HCC implantation murine model was used and the growth of hepatic tumours derived from luciferase-tagged MHCC97L cells were monitored. Five-week oral administration of mangiferin (50 mg/kg/2 days) decelerated hepatic tumour growth. Tumour regression was apparent after 3 weeks of mangiferin treatment (Fig. 1A) and continued through to completion of treatment (Fig. 1B). In addition, body weight remained similar between the two groups throughout the experimental period, suggesting that mangiferin was not toxic to the mice (Fig. 1C). Further histological analyses showed that control mice had irregular and invasive growth of the hepatic tumours. The elusive border between hepatic and tumour tissue, with tumour cells infiltrating the vascular lumen suggested rapid and invasive growth of HCC. In contrast, the tumours in the mangiferin group displayed a non-invasive growth pattern with a distinct border between the hepatic and tumour tissue, suggesting that the cells had less invasive properties after mangiferin treatment (Fig. 1D). The average vessel area and proliferative populations, as evidenced by CD31 and Ki67, respectively, were significantly smaller in the mangiferin group than in the control group (Fig. 1E). In sum, mangiferin administration suppressed orthotopic hepatic tumour growth in vivo.

Observing the significant inhibition of orthotopic tumour growth by mangiferin, we further investigated the inhibitory role of mangiferin on various functions in HCC cells. Two human hepatocellular carcinoma cell lines, MHCC97L and HLF were used. Cytotoxicity assay revealed that mangiferin induced no significant toxicity (more than 75% of cell viability) in HCC cells (Fig. 2A). However, 48 h of incubation with mangiferin at doses above 200 µg/mL led to gradual death of HLF but not MHCC97L cells. Although mangiferin at doses under 200 µg/mL were not cytotoxic to HCC cells, a dose of 120 µg/mL reduced cell proliferation, as determined by BrdU incorporation (Fig. 2B). Cell cycle analysis further confirmed the delay of the G1/S transition in HCC cells subject to mangiferin treatment, in a dose-dependent manner (Fig. 2C). These results suggest that mangiferin might defer HCC cell proliferation rate by halting the cells in G1 phase, without affecting cell viability. Consistently, clonogenic assays also showed reduced colony formation in MHCC97L and HLF cells following treatment with 120 µg/mL of mangiferin for 14 days (Fig. 2D). In sum, these results indicate that mangiferin decreases the reproductive capacity of HCC cells. Surprisingly mangiferin also reduced the migratory and chemotactic properties of HCC cells in a dose-dependent manner, as evidenced from the wound healing (Fig. 3A & 3B) and trans-well assays (Fig. 3C). In sum, these results demonstrated that mangiferin may interrupt HCC cell growth through various mechanisms.

To better understand how mangiferin inhibits HCC growth, we analysed the cellular action of mangiferin using PCR microarrays targeting major oncogenic and suppressive proteins in HCC (Fig. 4A). Applying a fold-change of 2.5 as a cut-off, we noted that, upon mangiferin treatment, 33 genes were down-regulated and 9 were up-regulated in MHCC97L cells (Fig. 4B and Table 2). Gene ontology (GO) analysis of expression profiles indicated that mangiferin treatment but not the control vehicles was associated with a significant enrichment of genes involved in Wnt pathway regulation (Fig. 4C). Noting that mangiferin potentially reduced proliferation-related genes in MHCC97L cells, we used quantitative PCR to further validate the mRNA expression of six related genes in orthotopic hepatic tumour tissues. The expression levels of these genes were reduced in tumours from mice treated with mangiferin but not in tumours from mice treated with the control vehicle (Fig. 4D). This finding suggests a potent inhibitory effect of mangiferin on tumour proliferation. Of the 6 deregulated genes, the expression of LEF1 was the most significantly suppressed. A similar downregulation profile of LEF1 mRNA was also observed in HLF cells treated with mangiferin (Fig. 4E).

Subsequently, we sought to analyse the clinical relevance of LEF1 in HCC patients using 40 cases of paired tumorous and non-tumorous samples extracted from the TCGA liver cancer transcriptome dataset. The expression of LEF1 was significantly higher in tumorous hepatic tissue than in non-tumour hepatic tissue (Fig. 4F). Kaplan-Meier analysis indicated that patients with high expression of LEF1 had significantly lower overall survival rates than patients with low LEF1 expression (Fig. 4G). In addition, increased LEF1 expression was associated with advanced pathological grade, according to the grading criteria established by the American Joint Committee of Cancer (AJCC) (Fig. 4H). This finding suggests that LEF1 expression may be a negative prognostic indicator for HCC patients. Overall, these results indicate that downregulation of LEF1—a critical gene correlated to HCC patient survival— by mangiferin, contributes to its inhibitory effect on HCC progression.

LEF1 expression is a prerequisite of the Wnt signalling pathway [28]. To further examine whether LEF1 suppression by mangiferin deactivated Wnt signalling, we quantified the transcriptional products of the Wnt pathway in mangiferin-treated HCC cells. Significant down-regulation of LEF1 trans-activated genes, MYC, axin2, MMP2 and CCND1 was observed in MHCC97L and HLF cells after mangiferin treatment (Fig. 5A). In line with these findings, nuclear localization of LEF1 in MHCC97L cells was blocked following mangiferin treatment (Fig. 5B). Immunoblot analysis also revealed that mangiferin treatment led to a dose-dependent reduction in nuclear LEF1 in MHCC97L and HLF cells (Fig. 5C). To further assess the role of LEF1 in mangiferin-mediated alterations in Wnt-related target gene expression, we transduced MHCC97L and HLF cells with a retroviral vector expressing pBABE-LEF1 [29]. The gain-of-function of LEF1 diminished Wnt gene repression by mangiferin (Fig. 5D). Furthermore, the differences between control- and mangiferin-treated cells regarding expression levels of Wnt-trans-activated genes were diminished by LEF1-overexpression (induced by transfection with pBABE-LEF1) but not by unaltered LEF1 expression (after transfection with an empty vector) (Fig. 5E). Consistent with the functional significance of increased Wnt-target genes, the inhibitory effect of mangiferin on cellular expansion was abolished by LEF1 overexpression (Fig. 5F). All these results further support the essential role of LEF1 inhibition in mediating suppression of Wnt signalling by mangiferin.

The activation of the canonical Wnt pathway involves stabilization of free β-catenin, followed by transcriptional activation of LEF1 through nuclear binding with β-catenin and initiation of Wnt-regulated downstream genes [11]. To further determine whether the inhibition of the Wnt pathway by mangiferin in HCC cells relies on β-catenin, we used recombinant protein Wnt3a to specifically activate β-catenin signalling. Addition of Wnt3a merely triggered the activation of LEF1. In addition, mangiferin treatment of HCC cells containing Wnt3a led to LEF1 reduction similar to that observed in cells without Wnt3a (Fig. 6A). These results suggest that β-catenin is not involved in LEF1 expression and the subsequent deactivation of Wnt signalling. This was further supported by the observation that transcriptional and translational activation of β-catenin remained unaltered in mangiferin-treated HCC cells upon Wnt stimulation (Fig. 6B & 6C).

The Wilms’ tumour 1 gene (WT1) is overexpressed in hepatic cancer cells and promotes cancer cell dedifferentiation and drug resistance [30]. We found that WT1 was down-regulated at both transcriptional and translational levels in mangiferin-treated cells (Fig. 6D & 6E). Similar results were also observed in mangiferin-treated hepatic tissue (Fig. 6F). We therefore hypothesized that expression of LEF1 is associated with WT1 expression in HCC cells. By analysing 244 pairs of data from the TCGA database, we found that WT1 and LEF1 mRNA expression were positively correlated (Fig. 6G), indicating that LEF1 expression might be regulated by WT1. To further examine the role of WT1 in regulating LEF1 gene expression, MHCC97L cells were transfected with the pAd/WT1-IRES-nAmCyan vector, targeting WT1. Accordingly, in the absence of mangiferin, WT1 overexpression led to increased expression of LEF1, indicating the transcriptional activation of LEF1 by WT1 (Fig. 6H). Our findings suggest that mangiferin repressed transcriptional activation of LEF1 in a WT1-dependent but β-catenin-independent manner.

A recent genome-wide screen revealed that WT1 has multiple targets in cells, including several proteins involved in the Wnt pathway [31], including LEF1. Another study of taste cells showed that knockdown of WT1 reduced the expression of LEF1 [32], which was in accordance with our observations in described in the preceding section. To determine the mechanism underlying transcriptional activation of LEF1 by WT1, we performed ChIP assays with antibodies to WT1. We detected the promoter region of LEF1 in WT1 antibody-precipitated nucleotides, which indicated that WT1 may activate LEF1 transcription through direct binding to its promoter region. More importantly, the occupancy of WT1 on the LEF1 promoter site was significantly reduced following mangiferin treatment (Fig. 7A). Sanger sequencing analysis further confirmed that the WT1 binding site was located in the LEF1 promoter region (Fig. 7B). To further investigate possible interactions between mangiferin and WT1, we performed in silico molecular docking analysis of mangiferin with the different binding sites of WT1. The mangiferin structure could strongly bind to the DNA binding site of WT1, according to a docking score of 5.722 pKd (Fig. 7C). Overall, our findings together with others reports, unveiled WT1 as a transcription factor of LEF1, enhancing its expression by binding to the promoter region. Also, mangiferin suppressed LEF1 expression by decreasing binding of WT1 to the LEF1 promoter region.

Several studies in the last 5 years have demonstrated an inhibitory effect of mangiferin on the growth and proliferation of myriad types of cancer cells, including cells from breast adenocarcinoma [33], prostate cancer [34], leukaemia [35], glioma [36], and nasopharyngeal cancer [37]. Yet, all these experiments were implemented at the cellular level and in vitro cellular models may not mimic the clinical cancer pathogenesis. Tumorigenesis is a complicated process involving interactions between tumour cells and immune cells, the extracellular matrix, and stem cell niches. These interactions may not all be modelled accurately in vitro. A recent study demonstrated a chemo preventive effect of mangiferin in a breast cancer xenograft model [23]. Mangiferin administered 1 week before subcutaneous xenotransplantation of cancer cells diminished tumour growth. This effect was considered a result of the deregulation of Wnt/β-catenin signalling. In our study, to systematically investigate the role of mangiferin on HCC growth, we used an orthotopic HCC implantation model that represents well the critical tumorigenesis events involved in cancer cell invasion, whereby hepatocytes are replaced by neoplastic cells and tumour neovascularization occurs. We found that treatment with mangiferin, in comparison with the control vehicle, was associated with a less invasive and less proliferative pattern upon tissue examination. Additional molecular analyses further confirmed the inhibitory effect of mangiferin on HCC cell proliferation and partially elucidated a mechanism of action. Specifically, the Wnt pathway was inhibited by mangiferin independent of changes to the intracellular level of β-catenin. Instead, LEF1, the direct executor of Wnt signalling, was prominently suppressed. This suggested that mangiferin may target multiple effector molecules in the Wnt pathway in inhibiting the growth of different types of cancers. Given that clinical and investigational treatments that target individual molecules are often compromised by the frequent mutation of cancer cells [38], the multi-target property of mangiferin may indicate its potential as a promising Wnt inhibitor for cancer treatment.

Classical activation of the Wnt pathway is initiated by the stabilization of β-catenin as well as its translocation from the cellular membrane to intracellular regions upon ligand association with Wnt receptors [39]. Translocation of β-catenin promotes LEF1 activation by forcing it to disengage from Groucho proteins, so that transcription activity of LEF1 is executed [11]. However, several studies have proposed that LEF1 plays a β-catenin-independent role in the Wnt signalling [17, 18, 40]. Grumolato et al. reported the LEF1-dependent activation of Wnt signalling was observed without the association of β-catenin with LEF1 [17, 40]. TCF1 and LEF1 mutants that lack β-catenin binding sites also physically interact with transcriptional factor ATF2 for Wnt signalling activation [40]. Other studies revealed that factors like ALY and Smad may cooperatively act on LEF1 to activate Wnt signalling in a β-catenin-independent manner [17, 18]. In our study, we observed the direct transcriptional repression of LEF1 by mangiferin regardless of the initiation of upstream Wnt/β-catenin through ligand binding. A previous study stated that Wnt ligand activation caused elevated LEF1 expression in a β-catenin-independent and LEF1-dependent manner [40], which supports our finding of β-catenin-independent LEF1 repression by mangiferin in HCC cells. LEF1 related signalling is a multi-targeted regulatory mechanism in which the aberrant activation of LEF1 results in cancer development by promoting processes such as invasion, proliferation, and self-renewal of cancer cells [28, 41, 42]. Our molecular finding that mangiferin represses LEF1 target genes accords with these previous studies, especially given our cellular finding mangiferin restricted HCC cell proliferation, migration, and chemotactic properties.

Furthermore, we found that mangiferin-mediated LEF1 suppression was associated with WT1 in HCC cells. A previous study showed that WT1 is overexpressed in hepatic cancer cells and promote cancer cell growth and drug resistance [30]. Also, WT1 mutation has been widely detected in different types of cancer cells [43]. However, the role of WT1 as a tumour suppressor or promoter remains enigmatic [44]. In our study, gain-of-function targeting of WT1 up-regulated LEF1 expression in MHCC97L cells, suggesting that WT1 is a transcriptional activator of LEF1. This observation accords with the finding of Gao et al. that knockdown of WT1 reduced LEF1 expression [32]. Notably, we demonstrated that WT1, as a transcriptional activator, could directly bind to the promoter region of LEF1. Treatment with mangiferin reduced this association and therefore suppressed LEF1 expression and subsequent Wnt signal activation (Fig. 7D). A previous genome-wide screening revealed that the Wnt pathway is the putative target of WT1 [31], and our findings provide further insight into the mechanism by which WT1 connects to and regulates Wnt activity and its related signalling protein.

In conclusion, the current study showed the in vivo suppressive effect of mangiferin on HCC using an orthotopic implantation model. Cellular investigations demonstrated the multifaceted interruption of HCC cell growth by mangiferin through various mechanisms, including inhibition of HCC cell expansion and invasion. Furthermore, our mechanistic study indicated that the HCC inhibitory effect of mangiferin was mediated through transcriptional repression of LEF1. LEF1 suppression by mangiferin was evidenced by the observations of: (i) reduced LEF1 expression in both hepatic tumour tissues and HCC cells treated with mangiferin; (ii) downregulation of LEF1 trans-activated genes in MHCC97L and HLF cells, following mangiferin treatment; and (iii) overexpression of LEF1 diminished the downregulation of LEF1 trans-activated genes by mangiferin. Finally, the mangiferin-mediated down-regulation of LEF1 was independent of β-catenin but involved the disruption of the interaction between WT1 and the LEF1 promoter. Our study demonstrates LEF1 transcriptional up-regulation through association with WT1 protein, and subsequent Wnt activation and HCC growth. The study also indicates that mangiferin is a potential Wnt inhibitor that targets WT1/LEF1-mediated Wnt activation for HCC treatment.

The authors would like to express our gratitude to Mr. Keith Wong, Ms. Cindy Lee, Mr. Alex Shek and Faculty Core Facility for their technical support. This work was partially supported by the research council of the University of Hong Kong [project code: 104003422, 104004092 and 104004460]; Wong’s donation [project code: 200006276]; the donation of Gaia Family Trust, New Zealand [project code: 200007008]; Research Grant from Guizhou Bailing Pharmaceutical Company (Project Code: 26607830); the Research Grants Committee (RGC) of Hong Kong, HKSAR [project code: 766211 and 17152116] and Shenzhen Basic Research Program (Project code: JCYJ20140903112959964).

The authors have no competing financial interests to declare.