miR-491-5p, mediated by Foxi1, functions as a tumor suppressor by targeting Wnt3a/β-catenin signaling in the development of gastric cancer

Accumulated evidence has suggested that microRNAs (miRNAs) have an important role in tumor development and progression by regulating diverse signaling pathways. However, the precise role of miRNAs in gastric cancer (GC) has not been elucidated. In this study, we describe the function and regulation network of miR-491-5p in GC. miR-491-5p is frequently downregulated in GC tissues compared with adjacent non-cancerous tissues. Forced expression of miR-491-5p significantly inhibits proliferation and colony formation, and promotes apoptosis in GC cells. Through bioinformatic analysis and luciferase assays, we confirm that miR-491-5p targets Wnt3a. Silencing Wnt3a inhibits cell proliferation and induces apoptosis. Similarly, restoration of Wnt3a counteracts the effects of miR-491-5p expression. Moreover, bioinformatic and luciferase assays indicate that the expression of miR-491-5p is regulated by Foxi1, which binds to its promoter and activates miR-491-5p expression. In conclusion, to the best of our knowledge, our findings are the first to demonstrate that Foxi1 is a key player in the transcriptional control of miR-491-5p and that miR-491-5p acts as an anti-oncogene by targeting Wnt3a/β-catenin signaling in GC. Our study reveals that Foxi1/miR-491-5p/Wnt3a/β-catenin signaling is critical in the progression of GC. Targeting the pathway described in this study may open up new prospects to restrict the progression of GC.

Gastric cancer (GC) is the fourth most frequent malignant disease and the second most common cause of cancerrelated deaths worldwide. 1 GC is an aggressive tumor that is generally diagnosed at an advanced stage. Although much progress has been made in treatment, the combination of therapy strategies has not contributed to a significant impact on the clinical outcome of patients with advanced GC. 2 An increasing number of studies have focused on the exploration of molecular mechanisms underlying the development of GC, which may provide a foundation for novel therapeutic targets. 3 MicroRNAs (miRNAs) are a class of small non-coding nucleotides (18-25 nucleotides in length) that can negatively regulate target gene expression, and they function as tumor promoters or suppressors. These small molecules often bind to 3′-untranslated regions (3′-UTR) of messenger RNA (mRNA) by suppressing translation, or by degrading mRNA according to complete or incomplete complementarity. 4 Among these miRNAs, miR-491-5p, located in the fourth intron of focadhesin (FOCAD) has been reported to be deregulated in several human cancers. [5][6][7][8][9] miR-491-5p acts as a tumor suppressor by targeting JMJD2B in ERa-positive breast cancer. 10 miR-491-5p and its targeted gene GIT1 are considered biomarkers for the prognosis of oral squamous cell carcinoma. 11 miR-491-5p has been shown to inhibit metastasis, and epithelial to mesenchymal transition in hepatocellular carcinoma and esophageal cancer by suppressing matrix metalloproteinase (MMP) or targeting protein for xenopus kinesin-like protein 2 (TPX2), respectively. 12,13 Besides, upregulation of miR-491-5p suppressed proliferation and induced apoptosis of the human osteosarcoma cells. 14 Our present study shows that miR-491-5p is markedly decreased in GC tissues compared to non-cancerous adjacent tissues, suggesting that miR-491-5p may act as a tumor suppressor in GC. miRNAs function as tumor suppressors or oncogenes by directly targeting specific genes. 15 On the basis of the literature reviews and gene target prediction databases, including TargetScan (http://www.targetscan.org/), miRanda (http://www.microrna.org/), and miRBASE (http://www.mirbase.org/), we hypothesized that Wnt3a may be a potential target of miR-491-5p. The Wnt3a gene is an important member of the Wnt ligand family, which exerts its function by activating the canonical Wnt signaling pathway. 16 When Wnt signaling is activated, Wnt ligand binds to its receptor frizzled (Fz) and co-receptor lipoprotein receptor-related protein (LRP5/6). This binding boosts the stabilization of β-catenin in the cytoplasm, inducing it to translocate into the nucleus where it forms a ternary complex with T-cell-specific transcription factor (TCF)/lymphoid enhancer-binding factor (LEF) followed by transcriptional activation of target genes, such as cyclin D1 and c-myc. 17 Wnt3a is overexpressed in advanced prostate cancer cells, which activates transcription factor (ATF3)induced tumors. 18 Wnt3a is highly expressed in most breast cancer cell lines 19 and oral squamous cell carcinoma. 20 However, in mouse or human melanoma cell lines, Wnt3a overexpression prohibits proliferation in vitro and tumor growth in vivo. 21 Moreover, Wnt3a inhibits the growth and proliferation of the lung and lacrimal glands, 22 induces cell death of B-cell precursor acute lymphoblastic leukemia (B-ALL), and inhibits the proliferation of several B-ALL cell lines, 23 These findings suggests that Wnt3a may have varied effects on different cancer types. A previous study has shown that Wnt3a is highly expressed in human scirrhous gastric carcinoma 44As3 cells with highly metastatic derivatives. 24 The expression pattern of Wnt3a in GC tissue and the underlying molecular mechanisms remain to be further studied.
To the best of our knowledge, the upstream regulator of miR-491 remains to be clarified. It has been recognized that transcription factors, epigenetic modification, and the processing of pri-miR to mature miRNAs could play important roles in miRNA expression regulation. 25 Thus, the potential regulation of miR-491-5p by transcription factors needs to be investigated. Bioinformatic databases, including UCSC genome browser tool, VISTA (http://genome.lbl.gov/cgi-bin/WGRVis-taInput5.pl?cfg_dir = gp_r4099_169), and JASPAR database (http://jaspar.genereg.net/) was applied and indicated the presence of a Foxi1-binding region in the promoter of the miR-491 gene. Foxi1, also known as HFH3, belongs to the forkhead family, and the specific function of this gene has not yet been determined. However, it is possible that Foxi1 plays an important role in the development of the cochlea and vestibular duct as well as embryogenesis. [26][27][28] Thus, it needs to be clarified if Foxi1 mediates miR-491-5p expression and plays a role in the development of GC. The aim of the present study was to explore the function and underlying mechanism, including the upstream transcription factor and downstream target gene of miR-491-5p, in GC carcinogenesis. We provide evidence that Foxi1 mediates miR-491-5p and plays a crucial role in the regulation of proliferation and apoptosis of GC cells via Wnt3a/β-catenin signaling pathways. Our data were obtained based on gainand loss-of-function studies of Foxi1, miR-491-5p, and Wnt3a in vivo and in vitro. To our knowledge, this is the first study that directly illustrates a Foxi1/miR-491-5p/Wnt3a axis in GC.

Results
miR-491-5p inhibits cell proliferation of GC in vitro. We analyzed the expression pattern of miR-491-5p in 22 pairs of GC tissue and matched adjacent non-cancerous tissue samples using qRT-PCR. Compared to normal tissues, miR-491-5p was significantly downregulated in GC samples ( Figure 1a). This result was also validated in four GC cell lines. miR-491-5p expression was decreased in SGC-7901, MKN45, BGC823, and AGS cell lines compared to the GES-1 cell line (Figure 1b). To clarify the function of miR-491-5p in GC, MKN45 and SGC-7901 cells were selected for further analyses.
To explore the function and role of miR-491-5p, gain-and loss-of-function analyses were conducted. The expression level of miR-491-5p was markedly increased in the pre-miR-491-transfected cells compared to the control vectortransfected cells (Supplementary Figure S1A); the inhibitory effect was moderate due to the low expression of endogenous miR-491-5p in MKN45/SGC-7901 cells (data not shown). MTT and colony formation assays were applied, and the results showed that overexpression of miR-491-5p in MKN45 and SGC-7901 cells exerted an inhibitory effect on cell growth and colony formation after transfection (Figures 1c and d), while the miR-491-5p inhibitor exerted moderate adverse effects on GC cells that may be caused by the low expression level of miR-491-5p in MKN45/SGC-7901 cells (Figures 2a and b). The cell growth inhibitory effect of miR-491-5p was also performed by MTT assay in GES-1 cells (Supplementary Figure S1B). These data indicated that miR-491-5p may act as a tumor suppressor in GC.   Figure 1f). However, the miR-491-5p inhibitor exhibited a slight difference in cell apoptosis and no significant difference compared to cells transfected with the negative control, which may be due to the low expression level and low inhibition efficiency of miR-491-5p in MKN45/SGC-7901 cells (Figures 2c and d). We also evaluated the related inhibitory effect of miR-491-5p on apoptosis and cell cycle in GES-1 cells ( Supplementary Figures S1C and D).
To investigate the underlying mechanisms of miR-491-5p in apoptosis and cell cycle regulation, we measured the expression levels of apoptosis-and cell cycle-related proteins in GC cells. miR-491 transfection in MKN45/SGC-7901 cells downregulated pro-caspase 3, BCL-2, CDK2, and CCNA2, but upregulated active caspase 3 and cleaved PARP ( Figure 1g). In contrast, these protein expression levels exhibited opposite trends after miR-491-5p inhibitor transfection ( Figure 2e and Supplementary Figure S1E). These findings demonstrate that miR-491-5p is overexpressed in GC cells and inhibits cell growth by leading to cell cycle arrest and inducing apoptosis in GC cells.
Wnt3a is a direct functional target of miR-491-5p in GC. We searched for potential miR-491-5p target genes using computer-aided miRNA target prediction programs. Wnt3a was selected as a candidate because it has a potential miR-491-5p-binding site in its 3′-UTR. Furthermore, Wnt3a has been reported to be upregulated in human cancers, and involved in tumor growth and metastasis. 23,29 To determine if miR-491-5p directly targets Wnt3a, we subcloned 3′-UTR Wnt3a fragments including wild-type (Wnt3a-WT) and mutant (Wnt3a-MUT) miR-491-5p-binding was significantly reduced in miR-491-overexpressing HEK293 cells. As expected, miR-491-5p failed to inhibit the luciferase activity of Wnt3a-MUT pmirGLO-3′-UTR vector, indicating that miR-491-5p binds directly to the 3′-UTR of Wnt3a ( Figure 3b). In accordance with this result, the luciferase activity of TCF/ LEF luciferase reporter was significantly decreased in miR-491 transfection group compared with miR-negative control transfection group, but not in miR-491+Wnt3a vector co-transfection group (Figure 3c). The same luciferase assays were also conducted in MKN45/SGC-7901 cells (Supplementary Figures S2A and B). This result further confirmed that miR-491 regulates Wnt3a/β-catenin signaling pathway by targeting the 3′-UTR of Wnt3a in the development of GC. In addition, when pre-miR-491, the miR-491-5p inhibitor and related controls were transfected into MKN45 and SGC-7901 cells, miR-491-5p decreased the expression of Wnt3a and its downstream target -β-catenin, and upregulated Wnt3a and β-catenin protein levels were observed when miR-491-5p was knocked down (Figure 3d). The expression levels of Wnt3a were significantly increased in gastric tissues compared to matched adjacent non-cancerous tissue in 22 pairs of tissues. Western blot data also confirmed the increased expression of Wnt3a in GC tissues ( Figure 3e). Meanwhile, miR-491-5p levels were inversely correlated with Wnt3a expression by qRT-PCR assay ( Figure 3f). Taken together, our data demonstrated that Wnt3a is a direct target of miR-491-5p. We also examined Wnt3a expression levels by immunohistochemistry (IHC) using a tissue microarray of 92 pairs of GC and matched normal tissues. The clinicopathological features of the patients are summarized in Supplementary Table S1. Wnt3a was upregulated in 64.8% cancer tissues compared to normal tissues ( Figure 3g). Overexpression of Wnt3a was significantly associated with a higher rate of perineural invasion in GC patients (P = 0.022, Table 1).
Knockdown of Wnt3a reduces the progression of GC cells and overexpression of Wnt3a eliminates the effects of miR-491-5p. We knocked down Wnt3a expression by RNA interference (small interfering RNA (siRNA)) to confirm that Wnt3a is implicated in the antitumor effects of miR-491-5p. Our results showed that Wnt3a was knocked down by siRNA both at the mRNA and protein levels ( Figure 4a). Similar to miR-491-5p-overexpressing cells, downregulation of Wnt3a significantly inhibited proliferation and slightly inhibited colony formation in MKN45/SGC-7901 cells (Figures 4b and c). Moreover, knockdown of Wnt3a induced apoptosis in MKN45/SGC-7901 cells similar to the effect of miR-491-5p overexpression ( Figure 4d). However, the influence of Wnt3a siRNA on cell cycle was not similar to miR-491-5p upregulation ( Figure 4e). According to western blot analysis, Wnt3a siRNA decreased the expression of β-catenin, pro-caspase 3, BCL-2, and CCND1, and it increased the expression of active caspase 3 and cleaved PARP ( Figure 4f). These findings indicated that miR-491-5p can regulate GC progression by directly targeting Wnt3a.
To further demonstrate that miR-491-5p inhibits tumor progression through targeting Wnt3a, we constructed rescue experiments. The expression levels of Wnt3a were markedly higher after transfection of the Wnt3a expression vector into   MiR-491-5p induces growth inhibition of SGC-7901 cells in vivo. To further validate the tumor growth suppression role of miR-491-5p, we performed xenograft assays. miR-491-expressing SGC-7901 cells and negative control cells were injected subcutaneously into lateral armpits of mice. As expected, miR-491-5p played an important role in inhibiting tumor growth in vivo. Tumor nodules derived from miR-491-5p-overexpressing cells grew substantially less than those of the control group (Figures 5a and b). qRT-PCR confirmed significantly increased levels of miR-491-5p in xenograft tumors overexpressing miR-491 ( Figure 5c). The tumor nodule weight was significantly less in miR-491treated tumors than that of negative control-treated tumors ( Figure 5d). On day 30, the average volume of miR-491-treated tumors was much less than that of control tumors ( Figure 5e). As expected, in vivo data showed that the expression of Wnt3a in tumor tissues was reduced in miR-491-treated tumors by western blot (Figure 5f). These findings were consistent with the in vitro results and indicated that miR-491-5p has an anti-growth ability in GC by targeting Wnt3a in vivo.
Foxi1 activates miR-491-5p expression. Public databases, including the GEO database, MERAV (http://merav.wi.mit. edu/), and cBioPortal (http://www.cbioportal.org/index.do), showed that Foxi1 is significantly decreased in GC tissues compared to non-tumor adjacent tissues (Supplementary Figure S1). In our present study, the Foxi1 mRNA level was markedly decreased in GC tissues and cell lines (Figures 6a  and b). The UCSC genome browser tool, VISTA, and the JASPAR database were applied, and we found a putative Foxi1-binding site located 0.2 kb upstream of the FOCAD gene (Figure 6c). When the Foxi1-binding site reporter constructs (including Foxi1-binding site-WT and Foxi1binding site-MUT) and Foxi1 expression vectors were co-transfected into HEK293 cells, the Foxi1-binding site-WT reporter had higher luciferase activity compared to the mutant reporter (Figure 6d). Consistent with these data, the ChIP experiment indicated the Foxi1 protein binds to the putative binding site upstream of miR-491 (Figures 6e  and f). The increased expression level of Foxi1 in  Figure S3D). Accordingly, overexpression of Foxi1 in GC cells led to increased miR-491-5p expression and decreased Wnt3a expression, suggesting there is an axis among Foxi1/ miR-491-5p/Wnt3a signaling (Figures 6g and h). In addition, Foxi1 inhibited cell proliferation, and induced apoptosis and cell cycle arrest in GC cells (Figures 6i-k). Furthermore, overexpression of Foxi1 decreased the expression levels of pro-caspase 3, BCL-2, CDK6, and CCND1, but increased the Foxi1/miR-491-5p/Wnt3a axis in gastric cancer R Sun et al expression level of active caspase 3 and cleaved PARP (Figure 6m), suggesting that Foxi1 contributes to the proliferation inhibition, apoptosis promotion, and cell cycle arrest by modulating miR-491-5p transcription in GC cells.

Discussion
The carcinogenesis of GC is a multifactorial process including genetic and epigenetic events. Emerging evidence indicates that miRNAs are dysregulated in various cancers, and function as tumor suppressors or oncogenes. Numerous studies have characterized the miRNA signatures of GC, but the exact role of miRNA dysregulation in the pathogenesis of GC remains elusive. miR-491-5p was selected as our research subject. The miR-491-5p gene locates within the fourth intron of FOCAD gene. Loss of FOCAD plays a potential tumor suppressor role in breast cancer, colorectal cancer, and glioblastoma. [30][31][32] Recently, miR-491-5p has been reported to be involved in several cancer types. [10][11][12][13] Until now, no reports have investigated the expression pattern, function, and underlying mechanisms of miR-491-5p in GC. In the current study, miR-491-5p was markedly downregulated in GC tissues. Moreover, our results suggested that miR-491-5p may play an important role in inhibiting GC growth, inhibiting proliferation, disrupting the cell cycle, and inducing apoptosis in vitro and in vivo. In addition, miR-491-5p altered the expression of the apoptosis/cell cycle-related proteins including caspase 3, PARP, BCL-2, CDK2, and CCNA2. Our data provide a more comprehensive understanding of the tumor suppressor role of miR-491-5p during GC progression. miRNAs cause mRNA degradation or inhibition by interacting with the 3′-UTR of target gene mRNA. Using bioinformatic analyses and a dual-luciferase reporter assay, we demonstrated that miR-491-5p directly targets Wnt3a by binding its 3′-UTR and inhibiting translation. In addition, our results showed that upregulation of miR-491-5p inhibited the expression of Wnt3a in MKN45/SGC-7901 cancer cells. Wnt3a exerts its function by activating the canonical Wnt signaling pathway followed by transcriptional activation of target genes, such as cyclin D1, C-myc, BCL-2, and survivin. Wnt3a is overexpressed in several cancer types; 18-20 a previous study has shown that Wnt3a is highly expressed in human scirrhous gastric carcinoma 44As3 cells with highly metastatic derivatives. 24 However, to the best of our knowledge, the function of Wnt3a and its clinicopathological significance in human GC has not yet been characterized. Consistent with previous data, our results showed that Wnt3a is upregulated in cancer tissues and downregulated in normal tissues. 24 To further clarify the tumor suppressor role of miR-491-5p by targeting Wnt3a, siRNA was used to knock down the expression of Wnt3a. Our data showed that Wnt3a silencing inhibited cell proliferation and induced cell apoptosis, which is similar to the effect of miR-491-5p overexpression in GC cells in vitro. Moreover, overexpression of Wnt3a substantially reversed the tumor-suppressive effects of miR-491-5p on cell proliferation and apoptosis in a rescue experiment. While, the colony-forming ability and cell cycle distribution were slightly altered by siWnt3a. Accordingly, the expression levels of related proteins, including β-catenin, pro-caspase 3, active caspase 3, cleaved PARP, BCL-2, and CCND1, were also altered by siWnt3a. In addition, overexpression of miR-491-5p by lentiviral infection also inhibited tumor growth in nude mice. The expression levels of Wnt3a were downregulated in tumor nodules injected by miR-491 lentivirus-infected cells compared to negative control lentivirus-infected cells, which was similar to the results in GC cell lines. Such studies have also been performed for lung cancer. 6 Together, these findings suggest that miR-491-5p mechanistically acts as a tumor suppressor via regulation of the Wnt3a/β-catenin signaling pathway.
In the current study, bioinformatic analyses suggested that Foxi1 might target the upstream region of miR-491. Furthermore, the GEO database, MERAV, cBioPortal database, and our present study show that the expression of Foxi1 is markedly decreased in GC tissues compared to their normal counterparts. These findings suggest that Foxi1 might be involved in the regulation of miR-491-5p in GC development. In our present study, function analyses showed that Foxi1 inhibited GC cell growth and proliferation. In addition, Foxi1 led to cell cycle arrest and induced cell apoptosis. Moreover, related proteins, including pro-caspase 3, active caspase 3, cleaved PARP, BCL-2, CCND1, and CDK6, were also changed by Foxi1 expression. On the basis of ChIP and luciferase assays, our results confirmed that Foxi1 directly binds to the miR-491 promoter. To the best of our knowledge, this is the first report demonstrating that Foxi1 may function as a tumor suppressor in GC development. Foxi1 belongs to the forkhead-box (FOX) family of genes, which have been  33 The FOXA1 gene is amplified and overexpressed in lung and esophageal cancer, and it is upregulated in pancreatic cancer and basal cell carcinoma. The FOXOl gene is fused to the PAX3 or PAX7 gene in rhabdomyosarcoma. 33 Our data provide new evidence for the carcinogenic mechanism of the FOX family.
There are several new points in the present study. First, our study is the first to report that overexpression of Wnt3a is associated with perineural invasion in GC patients. The underlying mechanism may be due to the activation of Wnt/ β-catenin signaling downstream of target MMPs. 34 Previously, it has been suggested that perineural invasion is a poor prognostic marker of GC. 35,36 A larger population of samples should be used to confirm if Wnt3a can serve as a predictive marker of GC. Second, this is the first study to report the tumor suppressor role of Foxi1 by transcriptional regulation of miR-491-5p in the progression of GC. Further studies are required to elucidate the role of Foxi1 in the carcinogenic mechanism by gene amplification, chromosomal translocation, or retroviral integration similar to other FOX members.
In conclusion, our data reported the following new findings: (1) miR-491-5p is frequently downregulated in GC tissues and cell lines; (2) miR-491-5p functions as a tumor suppressor in GC cells; (3) Wnt3a is a downstream target gene of miR-491-5p; (4) miR-491-5p inhibits cell proliferation and promotes apoptosis in GC cells by regulating Wnt3a/β-catenin signaling; (5) Foxi1 is significantly decreased in GC tissues and acts as a tumor suppressor; (6) Foxi1 binds to the miR-491 promoter and activates its expression; and (7) there is a regulation axis among Foxi1/miR-491-5p/Wnt3a/β-catenin signaling during the development of GC. Our study underscores the important role of miR-491-5p in GC progression, and we expect that our findings for the Foxi1/miR-491-5p/Wnt3a/β-catenin signaling axis will provide useful information for the development of more effective and promising therapies against GC.
Materials and Methods GC tissues and cancer-derived cell lines. In total, 44 gastric tissues were tested in this study, encompassing 22 tumor samples from GC patients and 22 matched adjacent non-malignant tissue samples from the same patients. All the tissues were surgically obtained from the First Affiliated Hospital of Xi'an Jiaotong University. No radiotherapy or chemotherapy was conducted before surgery, and the patients with prior cancer history were excluded from this study. All tissue specimens were snap-frozen in − 80°C after resection until needed. In addition, to investigate the expression and localization of Wnt3a, one tissue microarray including 92 patients with matched non-cancerous tissue was performed. All cases of GC were evaluated histologically by two senior pathologists independently according to Lauren's and the World Health Organization's classifications (IARC Press, Lyon, 2000) and staged using the TNM staging of the International Union Against Cancer (UICC, 2002). This study was approved by the Ethics Committee of Xi'an Jiaotong University, and informed consent was obtained from all individuals. Human GC cell lines, including BGC823 (poorly differentiated), SGC-7901 (poorly differentiated and metastatic), MKN45 (poorly differentiated), and AGS, in addition to the GES-1 non-malignant gastric epithelium cell line were maintained in the Key Laboratory of Environment and Genes Related to Diseases at Xi'an Jiaotong University. All the cells were cultured in RPMI-1640 medium (Thermo Scientific Hyclone, Waltham, MA, USA) supplemented with 10% fetal bovine serum at 37°C in a humidified cell incubator with an atmosphere of 5% CO 2 and were subcultured during their logarithmic phase.
RNA extraction and quantitative reverse transcription PCR. Total RNA was isolated from the tissues and cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to manufacturer's instructions. RNA was quantified using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Total RNA (500 ng) and PrimeScript RT reagent (Takara, Kusatsu, Japan) were used to generate cDNA, which was subjected to qRT-PCR using SYBR Green Master Mix (Takara) performed on a FTC-3000TM System (Funglyn Biotech Inc., Toronto, Ontario, Canada). β-actin and small RNA RNU6B (U6) were used as endogenous controls to normalize mRNA and miRNA, respectively. The 2 − ΔΔCT method was used to quantify the relative levels of mRNA and miR-491-5p. Each sample was performed in triplicate. All primers used in the present study are shown in Supplementary Table S2.
Expression vector construction. The miR-491 (pre-miR-491) expression vector and related control vector were constructed by synthetic oligonucleotides and cloned into the pcDNA6.2-GW/EmGFP vector (Invitrogen) between the EcoRI and HindIII sites as previously described. 37,38 Using bioinformatic analyses, we identified a fragment of Wnt3a as a miR-491-5p target. The 3′-UTR of the human Wnt3a mRNA was constructed by synthetic oligonucleotides and cloned into the pmir GLODual-Luciferase miRNA Target Expression Vector (Promega, Madison, WI, USA) between the SacI and XhoI sites as previously described. 37,38 A miR-491-5p inhibitor, an inhibitor negative control, and siRNAs targeting Wnt3a were chemically synthesized (GenePharma, Shanghai, China). Wnt3a was cloned into the GV230 vector (Genechem, Shanghai, China) between the XhoI and KpnI sites. Foxi1 was cloned into PEGFP-C1 between the HindIII and BamHI sites (GENEWIZ, Suzhou, China). A 127 bp sequence containing the wild-type or mutant Foxi1-binding site was cloned into the pGL3 promoter vector between the KpnI and XhoI sites (GENEWIZ). All the sequence information is given in Supplementary Table S2.
Oligonucleotide and plasmid transfection. JetSI-ENDO transfection reagents (Polyplus-transfection, Strasbourg, France) were used for cell transfection according to manufacturer's instructions. The transfection efficiency was verified by quantitative real-time PCR. The controls were used under the same experimental conditions as the miRNA mimics, thus allowing a transfection efficiency assessment during each assay.  Colony formation assay. For the colony formation assay, MKN45 and SGC-7901 (2000 cells per well) were seeded into six-well plates 24 h after transfection. After 10 days of incubation, the plates were gently washed with phosphate-buffered saline (PBS), and the colonies were stained with 0.1% crystal violet for 30 min after fixation with methanol for 10 min. Excess dye was rinsed off by PBS. Images of the colonies were acquired, and the colonies were counted using computer software (Quantity One, Bio-Rad, Hercules, CA, USA).
Cell apoptosis assay. In brief, MKN45 and SGC-7901 cells were seeded into 12-well plates at a density of 1 × 10 6  Orthotropic tumor model in nude mice. Tumorigenicity assays in animals were conducted according to the guidelines of the Animal Care and Use Committee of Xi'an Jiaotong University. Four-week-old male nude mice (BALB/cnude) were used to examine tumorigenicity. In brief, SGC-7901 cells were transfected with LV-GFP-miR-491 or LV-GFP-miR-negative control lentiviral vectors (Genechem), and 1 × 10 6 miR-491 and miR-negative control cells were separately injected into each mouse armpit, thus generating tumor comparisons within each mouse. Tumor size was measured every 3 days. Eight weeks after injection, mice were killed and tumor nodules were weighed. The tumor volume (V) was calculated as follows: V = (a × b 2 )/2; where a and b are the long and short axes of the tumor nodule, respectively. For end point experiments, the animals were anesthetized by isoflurane as described previously, and the tumors in living mice were measured by observing GFP expression using a photobiology system (Xenogen IVIS Spectrum imaging system (Xenogen, Alameda, CA, USA)). The blots were scanned, and the band density was then quantified by Quantity One imaging software (Bio-Rad, Hercules, CA, USA).
Immunohistochemistry. IHC was performed as described previously. 39 Tissue microarray sections were deparaffinized with xylene and hydrated using an alcohol gradient. Endogenous peroxidase-blocking and antigen retrieval were performed sequentially. The sections were then incubated with polyclonal rabbit anti-Wnt3a (Bioss, diluted 1:100) followed by incubation with secondary antibody conjugated with HRP. Detection was conducted by 3,3′-diaminobenzidine and hematoxylin. The degree of immunostaining was evaluated by two independent pathologists who were blind to the clinical data of the patients. We used a scoring standard for Wnt3a protein expression, and both the percentage of positive cells and intensity were considered as previously described.
Dual-luciferase assay. HEK293 cells were seeded into 96-well plates (10 000 cells per well), and 100 ng of pmirGLO-Wnt3a-3′-UTR vectors, including wild-type or mutated miR-491-5p-binding sites, was co-transfected with 100 ng of pre-miR-491 expression vector into HEK293 cell lines, and pmirGLO vector was used as the control. In a second experiment, the Foxi1 expression vector (or empty vector) was co-transfected with wild-type or mutated Foxi1-binding site reporter constructs, and a blank pGL3 luciferase vector was used as a positive control. Twenty-four hours after transfection, firefly and Renilla luciferase activity was examined using the Dual-Luciferase Assay System (Promega). The relative expression of firefly luciferase activity was normalized to Renilla luciferase activity. Each assay was performed three times.
TCF/LEF reporter assay. The pGL4.49 (luc2P/TCF-LEF/Hygro; 0.1 μg per well) vector containing eight copies of a TCF-LEF response element that drives transcription of the luciferase reporter gene luc2P (Photinus pyralis) was co-transfected with either miR-control, miR-491, and miR-491+Wnt3a vectors into HEK293 cells in a 96-well plate. Forty-eight hours post transfection, firefly luciferase activity was examined using the ONE-Glo Luciferase Assay System (Promega). Luminescence was detected in a 1420-Multilabel counter. For each assay, five independent replicates were performed. Each assay was performed three times.
Chromatin immunoprecipitation assay. The binding of Foxi1 to the miR-491 promoter was evaluated using chromatin immunoprecipitation (ChIP) analysis. Briefly, protein/DNA complexes obtained from SGC-7901 cells were crosslinked using 1% formaldehyde (Bio-Rad, Hercules, CA, USA) for 15 min at room temperature, and the reactions were quenched by adding glycine (0.125 M) for 30 min. The cells were rinsed two times with 5 ml of PBS. Cells were collected, and the nuclei were resuspended in Mg-NI, Mg-NIXP40, Ca-NI (0.5 M EGTA addition), and lysis buffer (with protease inhibitors). The samples were then sonicated, and the chromatin was sheared into an average length of~100-500 bp. The insoluble material was removed by centrifugation at 14 000 × g. Of the DNA/protein complexes, 100 μl was used as input for ChIP. Foxi1 was immunoprecipitated from the supernatant using an anti-Foxi1 antibody (normal rabbit IgG antibodies were used as negative controls) at 4°C for 4 h and bound to protein G sepharose (Invitrogen) during a 2 h incubation at 4°C. The immunoprecipitates were then washed two times successively with the following solutions: ChIP lysis buffer, ChIP lysis buffer containing 500 mM NaCl, LiCl/detergent solution (10 mM Tris-HCl, pH 8.0; 250 mM LiCl, 0.5% NP-40, 0.5% sodium deoxycholate, and 1 mM EDTA), and finally TE buffer (10 mM Tris and 1 mM EDTA, pH 8.0). A solution containing 1% SDS and 0.1 M sodium bicarbonate was used to elute the bound proteins from beads. The input and the eluent samples were reverse cross-linked with proteinase K by incubating at 65°C for 8 h. Phenol/chloroform (Invitrogen) was used to isolate the DNA from the samples followed by ethanol precipitation. Promoter binding was determined by PCR with primers spanning the miR-49 upstream regions and SYBR Green Master Mix (Takara) using a LightCycler 480 (Roche Diagnostics, Tokyo, Japan). Enrichments were calculated as percentage of the input. The primer sequence oligonucleotides used for PCR amplification are shown in Supplementary Table S2.
Statistical analysis. All data are presented as mean ± S.D. of at least three independent experiments. Differences between two groups or among three groups were analyzed by Student's t-test or one-way ANOVA. The linear correlation coefficient (Pearson r) was calculated to estimate the correlation between miR-491-5p values and Wnt3a levels in the matched GC tissues. The relationship between Wnt3a expression level and clinical parameters was calculated by the χ 2 -test. Data were considered to be statistically significant when P-values were o0.05. All of the Foxi1/miR-491-5p/Wnt3a axis in gastric cancer R Sun et al