MicroRNA-216a/217-induced epithelial-mesenchymal transition targets PTEN and SMAD7 to promote drug resistance and recurrence of liver cancer

Authors

  • Hongping Xia,

    1. Laboratory of Cancer Genomics, Division of Cellular and Molecular Research, Humphrey Oei Institute of Cancer Research, National Cancer Center, Singapore, Singapore
    Search for more papers by this author
  • London Lucien P.J. Ooi,

    1. Department of Surgical Oncology, National Cancer Center, Singapore, Singapore
    2. Department of General Surgery, Singapore General Hospital, Singapore, Singapore
    Search for more papers by this author
  • Kam M. Hui

    Corresponding author
    1. Laboratory of Cancer Genomics, Division of Cellular and Molecular Research, Humphrey Oei Institute of Cancer Research, National Cancer Center, Singapore, Singapore
    2. Cancer and Stem Cell Biology Program, Duke–National University of Singapore Graduate Medical School, Singapore, Singapore
    3. Department of Biochemistry, Yong Loo Lin School of Medicine, National University of Singapore, Singapore
    4. Institute of Molecular and Cell Biology, Singapore, Singapore
    • Address reprint requests to: Kam M. Hui, Ph.D., Division of Cellular and Molecular Research, National Cancer Center, 11 Hospital Drive, Singapore 169610, Singapore. E-mail: cmrhkm@nccs.com.sg; fax: +65 6226-3843.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants from the SingHealth Foundation, the National Medical Research Council, the Biomedical Research Council of Singapore, and The Singapore Millennium Foundation. The authors thank the NCCS Tissue Repository for providing the tissue specimens for this study, Prof. S.S. Jiang from Sun Yat-sen University (Guangzhou, China) for providing the pLL3.7-miR-control vectors, and Addgene for the plasmids.

Abstract

Tumor recurrence and metastases are the major obstacles to improving the prognosis of patients with hepatocellular carcinoma (HCC). To identify novel risk factors associated with HCC recurrence and metastases, we have established a panel of recurrence-associated microRNAs (miRNAs) by comparing miRNA expression in recurrent and nonrecurrent human HCC tissue samples using microarrays (recurrence is defined as recurrent disease occurring within a 2-year time point of the original treatment). Among the panel, expression of the miR-216a/217 cluster was consistently and significantly up-regulated in HCC tissue samples and cell lines associated with early tumor recurrence, poor disease-free survival, and an epithelial-mesenchymal transition (EMT) phenotype. Stable overexpression of miR-216a/217-induced EMT increased the stem-like cell population, migration, and metastatic ability of epithelial HCC cells. Phosphatase and tensin homolog (PTEN) and mothers against decapentaplegic homolog 7 (SMAD7) were subsequently identified as two functional targets of miR-216a/217, and both PTEN and SMAD7 were down-regulated in HCC. Ectopic expression of PTEN or SMAD7 partially rescued miR-216a/217-mediated EMT, cell migration, and stem-like properties of HCC cells. Previously, SMAD7 was shown to be a transforming growth factor beta (TGF-β) type 1 receptor antagonist. Here, we further demonstrated that overexpression of miR-216a/217 acted as a positive feedback regulator for the TGF-β pathway and the canonical pathway involved in the activation of phosphoinositide 3-kinase/protein kinase K (PI3K/Akt) signaling in HCC cells. Additionally, activation of the TGF-β- and PI3K/Akt-signaling pathways in HCC cells resulted in an acquired resistance to sorafenib, whereas blocking activation of the TGF-β pathway overcame miR-216a/217-induced sorafenib resistance and prevented tumor metastases in HCC. Conclusion: Overexpression of miR-216a/217 activates the PI3K/Akt and TGF-β pathways by targeting PTEN and SMAD7, contributing to hepatocarcinogenesis and tumor recurrence in HCC. (Hepatology 2013;58:629–641)

Abbreviations
Akt

protein kinase B

ANOVA

analysis of variance

CSC

cancer stem cell

DFS

disease-free survival

EMT

epithelial-mesenchymal transition

EpCAM

epithelial cell adhesion molecule

HCC

hepatocellular carcinoma

IST

Instructional Systems Technology

JAK2

Janus kinase 2

MET

mesenchymal epithelial transition; miRNA, microRNA

mRNA

messenger RNA

MTS

methyl tetrazolium salt

P-Akt

phosphorylated Akt

PI3K

phosphoinositide 3-kinase

PTEN

phosphatase and tensin homolog

qRT-PCR

quantitative reverse-transcription polymerase chain reaction

SD

standard deviation

SMAD7

mothers against decapentaplegic homolog 7

TGF-β

transforming growth factor beta

TGFBR1

TGF-β receptor type 1

UTR

untranslated region

WT

wild type.

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer and the third-leading cause of deaths from cancer worldwide. Recurrent disease is one of the most serious challenges for managing patients with HCC.[1] Although hepatic resection is a well-accepted therapy for early-stage HCC, many patients develop tumor recurrence and this converts the situation to a dismal prognosis.[2] Coupled with the inherent high resistance of HCC to chemotherapeutic drugs, recurrent disease forms the main cause of death in long-term evaluations. The identification of predictive molecular factors for tumor recurrence and understanding the roles of these markers in the molecular genetic mechanisms underlying HCC tumor recurrence would lead to improved overall clinical management of patients with HCC.

MicroRNAs (miRNAs) are a class of small endogenously expressed noncoding RNAs. The ability of some miRNAs to function as tumor promoters (miR-30d, miR-151, and miR-210) or suppressors (miR-122, let-7g, miR-29b, miR-193b, miR-194, miR-139, and miR-124) in hepatocarcinogenesis have led to new insights into the molecular pathways involved in HCC.[3, 4] Up to 90% of all human cancers, including HCC, are carcinomas, which are cell growths that originate in epithelial cells. Epithelial-mesenchymal transition (EMT) converts epithelial cells into mesenchymal cells, a normal embryological process frequently implicated in cancer aggressiveness and metastases. miRNAs have been demonstrated to play important regulatory functions in EMT.[5] Cancer stem cells (CSCs) within tumors are a small subset of cells capable of both tumor initiation and sustaining tumor growth. CSCs possess the capacity to self-renew and maintain tumor-initiating capacity through differentiation into the heterogeneous lineages of cancer cells that comprise the whole tumor and also provide the resource for cells that cause tumor recurrence and drug resistance. Emerging evidence indicates that miRNA-mediated EMT may also regulate CSC properties and thus provide a new avenue in understanding the regulatory mechanisms of CSCs and a potential new paradigm for the treatment of tumor invasion and metastases.[6-8] However, the regulatory roles of miRNAs in epithelial-mesenchymal and cancer stem-like transitions contributing to early recurrent disease and drug resistance of HCC remain to be elucidated.

In this study, we identified early HCC recurrent disease to be associated with up-regulation of the miR-216a/217 cluster by comparing miRNA expression profiles of HCC liver tissue from patients with early-recurrent and nonrecurrent disease. Overexpression of miR-216a/217 acted as a positive feedback regulator for the transforming growth factor beta (TGF-β) pathway and the canonical pathway involved in activation of phosphoinositide 3-kinase/protein kinase B (PI3K/Akt) signaling by targeting mothers against decapentaplegic homolog 7 (SMAD7) and phosphatase and tensin homolog (PTEN) in HCC cells, contributing to tumor recurrence and resistance to sorafenib. The overall strategy of this study is illustrated in Fig. S1 of the Supporting Materials.

Materials and Methods

RNA Extraction and Microarray Analysis

Total RNA from tissue samples or cell lines was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA). Quality and quantity of isolated total RNA was assessed using the Agilent 2100 Bioanalyzer and NanoDrop ND-1000 Spectrophotometer (Agilent, Santa Clara, CA). miRNA and messenger RNA (mRNA) profiling was performed as previously described.[9] Details are provided in the Supporting Materials.

Real-Time Quantitative Reverse-Transcription Polymerase Chain Reaction

Real-time quantitative reverse-transcription polymerase chain reaction (qRT-PCR) was performed as previously described[10, 11] (and in the Supporting Materials), using primers listed in Supporting Table 1. The U6 or HPRT1 internal control was used as an endogenous control, and fold changes were calculated by relative quantification (2−ΔCt).[12]

In Vitro Cell Migration Assay

Cell migration was assessed using cell-culture inserts (BD Biosciences, Bedford, MA), according to the manufacterer's guidelines. Details are provided in the Supporting Materials.

Sphere Formation Assay

The sphere formation assay was performed as previously described[13] (and in the Supporting Materials).

Flow Cytometry

Flow cytomety analysis was performed as previously described[13] and detected using a FACSCanto II flow cytometer (BD Biosciences).

Western Blotting and Immunofluorescence Analysis

Antibodies (Abs) and the procedure are described in the Supporting Materials.

Luciferase Reporter Assay

The 3'-UTR (untranslated region) sequence of PTEN and SMAD7 are predicted to interact with miR-216a, and miR-217 or a mutated sequence within the predicted target sites was synthesized and inserted into the XbaI and FseI sites of the pGL3 control luciferase reporter vector (Promega, Madison, WI) (Supporting Fig. 3). The luciferase reporter assay was performed as previously described[11, 13] (and in the Supporting Materials).

Animal Studies

All experiments on mice were approved by the SingHealth Institutional Animal Care and Use Committee. Details of animal studies are provided in the Supporting Materials.

Survival and Statistical Analysis

Experimental data are presented as the mean ± standard deviation (SD). All statistical analyses were performed using analysis of variance (ANOVA) or a two-tailed Student t test with either GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA) or Partek Genomics Suite software (Partek Incorporated, St. Louis, MO). Survival curves were calculated using Kaplan-Meier's method. Differences were considered statistically significant when P values were less than 0.05.

Results

Up-Regulation of the miR-216a/217 Cluster in HCC Tissues Was Associated With Tumor Recurrence and Survival

Because there have only been a few reports on the expression of miRNAs and their relation to early recurrent disease in patients with HCC, we conducted comprehensive miRNA profiling of liver biopsies from HCC patients with early and nonrecurrent diseases to identify miRNAs that are relevant to early recurrent disease. Early recurrence was defined as a recurrence within 2 years after a curative resection. miRNA expression profiles of 30 cases of liver biopsies, including 10 samples from HCC patients who had early recurrent disease over the 24-month observation period and 10 from patients who did not have early recurrent disease, were compared to 10 histologically normal tissue samples using the GeneChip miRNA 2.0 Array (Affymetrix, Inc., Santa Clara, CA). A panel of miRNAs with significant differential expression between early and nonrecurrent HCC and histologically normal liver tissue was derived through a series of ANOVA contrasts, and these miRNAs were selected for further validation and functional characterization. Among these miRNAs, expression of the miR-216a/217 cluster was significantly up-regulated in biopsies from HCC patients associated with early recurrent disease. Expression of miR-216a and miR-217 was >20- and >16-fold, respectively, when compared between HCC and normal liver tissue. Furthermore, expression of miR-216a and miR-217 was >4- and >3-fold, respectively, between biopsies from HCC patients associated with early and nonrecurrent HCC (Fig. 1A). These results from microarray studies were subsequently validated by qRT-PCR analysis with an independent cohort consisting of 29 tissue biopsies of HCC patients associated with early recurrent disease, 21 samples of patients with nonrecurrent disease, 10 histologically normal samples from HCC patients, and histologically normal liver tissue of 5 colorectal cancer patients who had liver metastases resected, which were used as reference normal liver tissue. The results obtained were consistent with the previous observation that expression of miR-216a/217 was significantly up-regulated in biopsies from HCC patients associated with early recurrent disease (Fig. 1B,C). Using the average expression value obtained for miR-216a/217 of the 50 samples studied as the cut-off point for Fisher's exact test and Kaplan-Meier's plots, it was demonstrated that high miR-216a/217 expression was significantly associated with poor survival (Fig. 1D,E).

Figure 1.

The miR-216a/217 cluster is up-regulated in HCC tissue and cell lines and is associated with tumor recurrence and survival. (A) Summary of microarray analyses. Upper panel shows that miR-216a and miR-217 are significantly up-regulated in HCC tumor tissue (T), compared to histologically normal tissue (MN). Lower panel shows that both miR-216a and miR-217 are significantly up-regulated in HCC tissue samples of patients with early recurrent disease (T-R), compared to patients with nonrecurrent disease over the same duration (T-NR). (B and C) Validation of the expression of miR-216a (B) and miR-217 (C) by qRT-PCR. (D and E) Expression of miR-216a (D) and miR-217 (E) was associated with the DFS of patients with HCC. The average expression value obtained for miR-216a/217 of the 50 samples studied by qRT-PCR was chosen as the cut-off point for survival analysis using Kaplan-Meier's method.

The miR-216a/217 Cluster Was Associated With EMT Phenotype and Migration Ability in Human HCC Cell Lines

Next, we analyzed the expression of miR-216a/miR-217 in a panel of liver cancer cell lines by qRT-PCR. Compared to normal liver tissues, expression of miR-216a/217 was significantly up-regulated in all HCC cell lines studied (Fig. 2A,B). It was also observed that epithelial HCC cells, such as HepG2 and PLC/PRF/5, had high expression of E-cadherin and low expression of vimentin, whereas HCC cells with a mesenchymal phenotype, such as SNU-449 and HLE, demonstrated low expression of E-cadherin and high expression of vimentin (Fig. 2A,B and Supporting Fig. 2A,B). The data suggest that expression of the miR-216a/217 cluster may be associated with EMT in HCC.

Figure 2.

The miR-216a/217 cluster was associated with the EMT phenotype and migration ability of epithelial HCC cells. (A and B) Expression of miR-216a/217 was studied in a panel of liver cancer cell lines. Up-regulation of miR-216a (A) and miR-217 (B) was associated with their EMT phenotype: Expression of miR-216a/217 was higher in mesenchymal phenotypic cells (HCCLM3, Bel-7404, HLE, SK-HEP1, and SNU-449), compared with cells with a basal epithelial phenotype (HepG2, Hep3B, HuH7, and PLC/PRF/5) (P < 0.05). (C) HepG2 and PLC/PRF/5 cells stably transfected with the miR-216a/217 cluster were stained for E-cadherin, vimentin, and Hoechst 33342 and analyzed by confocal microscopy. Red signal represents staining for E-cadherin (upper panel) and vimentin (lower panel). Nuclear DNA was detected by staining with Hoechst 33342. Scale bar represents 20 μm. (D) Expression of the epithelial protein, E-cadherin, and the mesenchymal protein, vimentin, in miR-216a/217 stably transfected HepG2 and PLC/PRF/5 cells were detected by western blotting. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. (E) Transwell migration assays of HepG2-miR-216a/217 (upper panel) and PLC/PRF/5-miR-216a/217 cells (middle panel). Migrated cells were plotted as the average number of cells per field of view from three different experiments (lower panel), as described in Materials and Methods. Original magnification: ×40. Error bars represent SD, and asterisks denote statistical significance <0.05.

Because miR-216a and miR-217 are clustered miRNAs located on human chromosome 2, we then transfected two HCC cell lines (HepG2 and PLC/PRF/5) with more epithelial phenotypes and relatively low expression of miR-216a/217 with either pLL3.7-Pre-miR-216a/217 (P-miR-216a/217) or pLL3.7-miR-control vector (P-miR-control). Stable cell lines overexpressing both miR-216a and miR-217 together were established and were tentatively named as HepG2-miR-216a/217 or PLC/PRF/5-miR-216a/217. Expression of miR-216a and miR-217 in these cells was confirmed by qRT-PCR (Supporting Fig. 2C,D). Compared to P-miR-control transfected cells, up-regulation of miR-216a/217 was associated with the observed dramatic morphological changes of HepG2-miR-216a/217 and PLC/PRF/5-miR-216a/217 cells from an epithelial cobblestone phenotype to an elongated fibroblastic phenotype, which is indicative of EMT (Supporting Fig. 2E). Induction of EMT in HepG2-miR-216a/217 and PLC/PRF/5-miR-216a/217 cells was also associated with reduced expression of E-cadherin and an elevated expression of vimentin (Fig. 2C,D). EMT has been indicated as a key step in initiating cancer cell migration.[14] Hence, the migration potential of HepG2-miR-216a/217 and PLC/PRF/5-miR-216a/217 cells was examined using the transwell migration assay in vitro. Results indicated that stable overexpression of the miR-216a/217 cluster significantly promoted the migration ability of HepG2-miR-216a/217 and PLC/PRF/5-miR-216a/217 cells in vitro (Fig. 2E). These data indicated that overexpression of miR-216a/217 in HCC with epithelial phenotypes induced EMT and enhanced migration abilities.

Next, HLE cells with a mesenchymal phenotype were used as recipient cells for transfection of antagomir-miR-216a/217 (Genepharma, Shanghai, China). (Antagomirs, also known as anti-miRs or blockmirs, are a novel class of chemically engineered oligonucleotides used to silence endogenous miRNAs.) After the silencing of miR-216a/217 (Supporting Fig. 3A), striking morphological changes consistent with those of mesenchymal-to-epithelial transition (MET) were observed (Supporting Fig. 3B). Up-regulation of E-cadherin, an epithelial biomarker, and reduced expression of vimentin, a mesenchymal biomarker, were also observed (Supporting Fig. 3C). Furthermore, we also examined the expression of miR-216a/217 on the proliferation and apoptosis of liver cancer cells. A significant increase in cell proliferation was observed in PLC/PRF/5 at 72 hours after transfection of the p-miR-216a/217-overexpressing vector, whereas transfection of the antagomir-miR-216a/217 into HLE cells significantly decreased cell proliferation (Supporting Fig. 4A). The number of apoptotic cells (Annexin V+ cells) was not significantly affected in PLC/PRF/5 and HLE cells by modulating expression of the miR-216a/217 cluster (Supporting Fig. 4B).

Stable Overexpression of miR-216a/217 Increases Stem-Like Properties and Promotes Tumor Growth and Metastases of HCC Cells

The recent discovery of the emergence of CSCs occurred, in part, as a result of miRNA-mediated EMT, which has provided a new avenue in understanding the regulatory mechanisms in CSCs and drug resistance. Because specific CSC markers have not been well defined for most CSCs, sphere-forming ability has emerged as a useful tool to evaluate the stemness characteristics of cells and for the enhanced enrichment of potential CSCs. Therefore, we evaluated HepG2-miR-216a/217 and PLC/PRF/5-miR-216a/217 cells for their ability to form tumor spheres. It was observed that HepG2-miR-216a/217 and PLC/PRF/5-miR-216a/217 generated 2∼3-fold more spheres than corresponding control cells (Fig. 3A,B). Flow cytometric analysis further demonstrated that sphere-forming cells derived from HepG2-miR-216a/217 and PLC/PRF/5-miR-216a/217 cells gave an enriched epithelial cell adhesion molecule (EpCAM)+ cell subpopulation, consistent with reported characteristics of liver CSCs[15] (Fig. 3C,D). The parental HepG2 had a small percentage of EpCAM+ cell subpopulation (12.6%), which was increased to 23.9% after transfection with miR-216a/217 (Fig. 3C). This suggests that the miR-216a/217 cluster may play an important role in regulating the stem-like traits of HCC cells by inducing EMT. We further investigated the effects of miR-216a/217 on the tumor growth and metastatic potential of HCC cells using an orthotopic xenograft tumor model in vivo. Results indicated that stable overexpression of miR-216a/217 in the PLC/PRF/5-miR-216a/217 cells significantly promoted tumor growth in an orthotopic xenograft tumor model (Fig. 3E). More significantly, when lung tissues of mice were harvested at the end point of the experiments, all of the mice inoculated with PLC/PRF/5-miR-216a/217 cells gave good bioluminescent signals, indicating the presence of lung metastases (Fig. 3F). In contrast, no lung bioluminescent signals were detected in mice inoculated with PLC/PRF/5-P-miR-control cells (Fig. 3F). These data indicate that overexpression of miR-216a/217 increases stem-like properties and promotes tumor growth and metastases of epithelial HCC cells.

Figure 3.

Stable overexpression of miR-216a/217 increases stem-like properties and promotes tumor growth and metastases of epithelial HCC cells. (A and B) Stable overexpression of miR-216a/217 increased sphere formation of HCC cells. Single stable transfected HepG2 and PLC/PRF/5 cells were plated onto a six-well, ultralow attachment plate (Materials and Methods). After 2-3 weeks of culture, the number of spheres (diameter, >40 μm) was observed under a fluorescence microscope and counted in three randomly selected fields at the magnification of ×40. The results are summarized in the bar graph (B). Error bars represent SD, and asterisks denote statistical significance <0.05. (C and D) Flow cytometric analysis of EpCAM+ cells. Stable miR-216a/217-transfected HepG2-miR-216a/217 (C) and PLC/PRF/5-miR-216a/217 cells (D) exhibited an increase of EpCAM+ cells. (E and F) Stable overexpression of miR-216a/217 promotes tumor growth and metastases of HCC cells with a basal epithelial phenotype. Compared to mice inoculated with PLC/PRF/5-P-miR-control cells, stable overexpression of miR-216a/217 in the PLC/PRF/5-miR-216a/217 cells significantly promoted tumor growth in an orthotopic xenograft tumor model (E). (F) Images showing bioluminescent signals detected in lungs of mice injected with PLC/PRF/5-P-miR-control cells (upper panel) and PLC/PRF/5-miR-216a/217 cells when harvested at the end of the experiments (lower panel). Bioluminescent signals were measured every week, and the statistical significance of the bioluminescent signal in each group was evaluated using the Student t test.

SMAD7 and PTEN Are Identified as Two Functional Downstream Targets of miR-216a/217 and Both Genes Are Down-Regulated in HCC

To elucidate the molecular mechanisms by which the miR-216a/217 cluster induces EMT in HCC, we employed several computational algorithms to identify the potential functional targets for the miR-216a/217 cluster. Using miRecords, an integrated resource for microRNA-target interactions,[16] a panel of molecules were predicted to be potential targets of the miR-216a/217 cluster with six miRNA target prediction programs (Supporting Table 2). Previously, we established an expression database for HCC using Affymetrix Human Genome U133 plus 2.0 Arrays (Affymetrix).[9, 11] Expression of the predicted potential targets identified for the miR-216a/217 cluster was analyzed in our HCC expression database. It was identified that SMAD7 and Janus kinase 2 (JAK2) were significantly down-regulated in HCC, compared to adjacent histologically normal liver tissues (Supporting Fig. 5A,C). In comparison, expression of SMAD7, but not JAK2, in PLC/PRF/5-miR-216a/217 cells was significantly reduced (Fig. 4A). Previous reports demonstrated that PTEN is also a target of miR-216a/217.[17] PTEN was also significantly down-regulated in HCC, compared to adjacent histologically normal, liver tissue in our expression database for HCC (Supporting Fig. 5B). This prompted us to study the expression of PTEN in PLC/PRF/5-miR-216a/217 cells, revealing a significant down-regulation (Fig. 4A). The increased expression of SMAD7 and PTEN was also observed in mesenchymal phenotype HLE cells transfected with antagomir-miR-216a/217 (Supporting Fig. 3C).

Figure 4.

SMAD7 and PTEN are identified as two functional downstream targets of miR-216a/217 and are down-regulated in HCC samples. (A) Western blotting analysis of SMAD7 and PTEN expression in P-miR-control- and P-miR-216a/217-transfected PLC/PRF/5 cells. (B) Effect of cotransfection of P-miR-216a/217 with wild-type (wt) and mutant (mut) pGL3-PTEN and SMAD7 constructs (Supporting Fig. 3A-C) to PLC/PRF/5 cells by luciferase reporter assays. Data were normalized by the ratio of firefly and Renilla luciferase activities measured at 48 hours post-transfection. Bar graph shows mean ± SD in three independent transfection experiments. *P < 0.05. (C and D) Expression of SMAD7 and PTEN in the same 50 HCC tissue samples studied in Fig. 1B,C. (E and F) Expression of PTEN (E) and SMAD7 (F) was associated with the DFS of patients with HCC. The average expression value obtained for PTEN and SMAD7 of the 50 samples studied by qRT-PCR was chosen as the cut-off point for survival analysis using Kaplan-Meier's method.

To further demonstrate that SMAD7 and PTEN are directly targeted by miR-216a/217 in HCC cells, we investigated whether the miR-216a/217 cluster directly interacted with the 3'-UTR of SMAD7 and PTEN mRNA by a dual-luciferase reporter assay. The predicted 3'-UTR sequence of SMAD7 and PTEN that interacted with miR-216a/217, together with a corresponding mutated sequence within the predicted target sites, were synthesized and inserted into the XbaI and FseI sites of the pGL3 control vector (Promega) (Supporting Fig. 6A-D). These constructs were referred to as pGL3-SMAD7-3'UTR-wt and pGL3-SMAD7-3'UTR-mut, pGL3-PTEN-3'UTR-wt, and pGL3-PTEN-3'UTR-mut. Cotransfection of P-miR-216a/217 with the various pGL3-constructs to PLC/PRF/5 cells significantly suppressed luciferase activity noted only with the wild-type (WT) 3'-UTRs (Fig. 4B). The suppressive effect of P-miR-216a/217 was abrogated when mutated 3'-UTR pGL3-constructs were used, confirming SAMD7 and PTEN were indeed direct downstream functional targets of miR-216a/217.

Expression of SMAD7 and PTEN was further validated by qRT-PCR in the previous cohort of 50 HCC tissue biopsies, 10 histologically normal samples from HCC patients, and histologically normal liver tissue of 5 colorectal cancer patients who had liver metastases. Both PTEN and SMAD7 were demonstrated to be significantly down-regulated in HCC tissue, compared to adjacent histologically normal liver samples (P = 0.001 and P = 0.0012, respectively) and between HCC samples of HCC patients with early recurrent and nonrecurrent disease (P = 0.004 and P = 0.0014, respectively) (Fig. 4C,D). When the average expression value obtained for PTEN and SMAD7 of the 50 HCC samples studied was used as the cut-off point for Fisher's exact test and Kaplan-Meier's plots, it was demonstrated that low PTEN or SMAD7 expression was significantly associated with comparatively poorer survival (Fig. 4E,F). Therefore, overexpression of the miR-216a/217 cluster inhibits expression of SMAD7 and PTEN in HCC cells and correlates with early recurrence and survival of HCC disease.

Reexpression of SMAD7 or PTEN Partially Rescues miR-216a/217-Mediated EMT, Cell Migration, and CSC-Like Properties in HCC Cells

To further study the roles of SMAD7 and PTEN in miR-216a/217 cluster-mediated EMT, cell migration, and CSC-like properties in HCC cells, we rescued the expression of SMAD7 and PTEN in PLC/PRF/5-miR-216a/217 cells by transfecting the plasmids carrying WT SMAD7 (pCMV5-SMAD7) or PTEN (pcDNA3.1-PTEN) (Addgene, Cambridge, MA) into PLC/PRF/5-miR-216a/217 cells.[18, 19] Reexpression of either SMAD7 or PTEN in PLC/PRF/5-miR-216a/217 cells, as confirmed by western blotting analysis (Fig. 5A), induced a dramatic morphological change of PLC/PRF/5-miR-216a/217 cells (Fig. S6E), implicating EMT. Induction of EMT observed with pCMV5-SMAD7 or pcDNA3.1-PTEN in PLC/PRF/5-miR-216a/217 cells was associated with up-regulation of E-cadherin, an epithelial biomarker, and reduced expression of vimentin, a mesenchymal biomarker (Fig. 5A,B). Consistent with these results, the migratory ability of PLC/PRF/5-miR-216a/217 cells was partially rescued after transfection with pCMV5-SMAD7 or pcDNA3.1-PTEN (Fig. 5C), and the sphere-forming ability of PLC/PRF/5-miR-216a/217 cells was reduced by 2∼3-fold, compared to cells transfected with control plasmids (Fig. 5D). Furthermore, flow cytometric analysis also demonstrated that reexpression of SMAD7 or PTEN partially decreased the EpCAM+ cell subpopulation in transfected PLC/PRF/5-miR-216a/217 cells (Fig. 5E). All the data indicate that reexpression of SAMD7 or PTEN could partially rescue miR-216a/217-mediated EMT, cell migration, and stem-like properties in HCC cells.

Figure 5.

Reexpression of SMAD7 or PTEN partially rescued miR-216a/217-mediated EMT, cell migration, and stem-like properties in PLC/PRF/5-miR-216a/217 cells. (A) Western blotting analysis of the expression of the epithelial protein, E-cadherin, and the mesenchymal proteins, vimentin in SMAD7- or PTEN-transfected PLC/PRF/5-miR-216a/217 cells. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. (B) Expressions of E-cadherin, vimentin, and Hoechst 33342 after transfection of SMAD7 or PTEN into PLC/PRF/5-miR-216a/217 cells were analyzed by confocal microscopy. Red signal represents staining for E-cadherin (upper panel) and vimentin (lower panel). Nuclear DNA was detected by staining with Hoechst 33342. Scale bar represents 20 μm. (C) Migratory properties of the stable transfected cells following reexpression of SMAD7 or PTEN in PLC/PRF/5-miR-216a/217 cells were analyzed by transwell migration assays. Migrated cells were plotted as the average number of cells per field of view from three different experiments (lower panel), as described in Materials and Methods. Original magnification: ×40. Results are summarized in the bar graph. Error bars represent standard error of the mean, and asterisks denote statistical significance <0.05. (D) Sphere formation properties of stable cells after reexpression of SMAD7 or PTEN in PLC/PRF/5-miR-216a/217 cells were analyzed by sphere formation assays. Single stable transfected cells were plated onto a six-well ultralow attachment plate (Materials and Methods). After 2-3 weeks of culture, the number of spheres (diameter, >40 μm) was observed under a fluorescence microscope and counted in three randomly selected fields at a magnification of ×40. Results are summarized in the bar graph. (E) Flow cytometric analysis of the EpCAM+ cell population in the stable transfected cells after reexpression of SMAD7 or PTEN in PLC/PRF/5-miR-216a/217 cells.

Overexpression of miR-216a/217 Activates the TGF-β Pathway and Induces Resistance to Sorafenib by Targeting PI3K-Akt Signaling in HCC Cells

SMAD7 has been shown to be a TGF-β receptor type 1 (TGFBR1) antagonist.[18, 20] To evaluate whether overexpression of miR-216a/217 leading to down-regulation of SMAD7 may activate the TGF-β pathway, we studied the expression of TGFBR1 in PLC/PRF/5-miR-216a/217 cells. Consistent with this hypothesis, up-regulation of TGFBR1 was observed in PLC/PRF/5-miR-216a/217 cells (Fig. 6A). Additionally, and in corroboration with an earlier report demonstrating up-regulation of miR-216a/217 by TGF-β in mouse mesangial cells, we observed that TGF-β treatment also induced up-regulation of miR-216a/217 in HCC cells (Supporting Fig. 7A,B). This indicated that overexpression of miR-216a/217 targets SMAD7, which could, in turn, increase expression of the TGF-β pathway member, TGFBR1, in a positive feedback mechanism in HCC cells and prompted the suggestion that the TGF-β pathway could be activated by overexpression of miR-216a/217.

Figure 6.

Overexpression of miR-216a/217 activates the TGF-β pathway to induce sorafenib resistance by activation of PI3K/Akt signaling in HCC cells. (A) Western blotting analysis of the expression of TGFBR1, P-Akt, Snail, and β-catenin in PLC/PRF/5-miR-216a/217 cells. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. (B) Dose-effect analysis of PLC/PRF/5-miR-216a/217 and PLC/PRF/5-miR-control cells exposed to sorafenib as evaluated by MTS assay. Error bars represent SD, and asterisks denote statistical significance <0.05. (C) Western blotting analysis of the expression of TGFBR1 and P-Akt in dimethyl sulfoxide (DMSO)- or LY2109761-treated PLC/PRF/5-miR-216a/217 cells. GAPDH was used as a loading control. (D) Dose-effect analysis of PLC/PRF/5-miR-216a/217 cells exposed to sorafenib in the presence of DMSO or 1 μg/μL of LY2109761 as evaluated by MTS assays. (E and F) PLC/PRF/5-miR-216a/217 were implanted into livers of recipient BALB/c nude mice and treated with sorafenib alone or sorafenib and LY2109761. The combination of sorafenib and LY2109761 overcame miR-216a/217-mediated sorafenib resistance and metastases of HCC cells (E). (F) Images showing bioluminescent signal detected in lungs of mice injected with PLC/PRF/5-P-miR-control cells and treated with sorafenib alone (upper panel) or sorafenib plus LY2109761 (lower panel) when harvested at the end of the experiments. Bioluminescent signals were measured every week, and the statistical significance of bioluminescent signals in each group was evaluated using the Student t test.

Because PI3K-Akt signaling can be activated by ablating PTEN expression,[6] we also tested whether the PI3K-Akt-signaling pathway was activated in PLC/PRF/5-miR-216a/217 cells. Akt exerts its effects in cells by phosphorylating a variety of downstream substrates. Compared to control cells, Akt phosphorylation (Ser473) was up-regulated in PLC/PRF/5-miR-216a/217 cells (Fig. 6A). When Kaplan-Meier's survival analysis between HCC patients with early recurrent and nonrecurrent disease was performed, a significant difference in disease-free survival (P < 0.0001) was observed (Supporting Fig. 8A). Immunohistochemical studies of expression of phosphorylated Akt (P-Akt) in matched normal, early-recurrent, and nonrecurrent HCC liver tissue samples also showed a significant difference in expression of P-Akt between early recurrent and nonrecurrent HCCs (Supporting Fig. 8B). In addition, phosphorylation of Akt downstream targets, including the mesenchymal and stemness markers, Snail and β-catenin,[21] was also up-regulated in PLC/PRF/5-miR-216a/217 cells (Fig. 6A). The data suggest that miR-216a/217-mediated EMT and stem-like features observed in early recurrent HCC disease may be associated with activation of the PTEN/PI3K/Akt pathway, which activates the downstream signal transduction pathways, including the Wnt/β-catenin pathway.

Previous studies have shown that activation of the PI3K/Akt-signaling pathway can confer resistance to sorafenib in HCC cells.[22] Therefore, we studied the sensitivity of PLC/PRF/5-miR-216a/217 to sorafenib. Methyl tetrazolium salt (MTS) assays demonstrated that PLC/PRF/5-miR-216a/217 cells were comparatively more resistant to sorafenib than PLC/PRF/5-miR-control cells (Fig. 6B). When PLC/PRF/5-miR-216a/217 cells were treated with LY2109761 (1 μM), a TGF-β type I/II receptor kinase inhibitor, strong inhibition of TGFBR1 expression and Akt phosphorylation in PLC/PRF/5-miR-216a/217 cells was detected (Fig. 6C). Treatment of LY2109761 (1 μM) also decreased expression of miR-216a and miR-217 in HepG2 and PLC/PRF/5 cells (Supporting Fig. 7C,D). To determine whether blocking the TGF-β pathway would overcome miR-216a/217-induced sorafenib resistance in HCC cells, we examined the effect of the presence or absence of LY2109761 (1 μM) on sensitivity of PLC/PRF/5-miR-216a/217 cells to sorafenib. MTS assays showed that a low dose of LY2109761 (1 μM) did not significantly inhibit cell growth (Supporting Fig. 6E), but was sufficient to overcome miR-216a/217-induced sorafenib resistance in PLC/PRF/5-miR-216a/217 cells (Fig. 6D). Using an orthotopic model of liver tumor established by surgical implantation of PLC/PRF/5-miR-216a/217 tumor cubes into the liver of the recipient mouse, it was demonstrated that sorafenib (20 mg/kg) plus LY2109761 (10 mg/kg) significantly inhibited tumor growth, when compared with sorafenib alone (Fig. 6E). More important, when bioluminescent signals from lung metastases were assessed to determine the metastatic ability of PLC/PRF/5-miR-216a/217 cells, it was observed that sorafenib plus LY2109761 significantly reduced mean bioluminescent signals of lung metastases, compared with treatment with sorafenib alone (Fig. 6F). The data further indicate that blocking the TGF-β pathway can overcome miR-216a/217-induced sorafenib resistance and tumor metastases in HCC.

Discussion

There are reports demonstrating the deregulation of miRNAs in HCC. However, miRNAs that play a specific role in the early recurrence and metastases of HCC have not been well documented.[3, 4] In this study, we demonstrated that the expression of miR-216a/217 was markedly increased in HCC tissue from patients with early recurrence. Furthermore, up-regulation of the miR-216a/217 cluster in a panel of liver cancer cells and HCC patients with early recurrent disease was associated with a more prominent EMT phenotype and poorer disease-free survival (DFS). These clinical observations corroborated well with the in vitro and in vivo findings reported in the present article using experimental animals and human HCC cell lines.

By examining the expression of HCC-related miRNAs between precancerous and cancerous liver tissues, an earlier study reported that miR-216a and miR-224 were significantly up-regulated in HCC tissues, and the elevation of miR-216a was mainly identified in male patients. To address the observed gender difference, these researchers showed that pri-miR-216a is activated transcriptionally by the androgen pathway in a ligand-dependent manner, and the TSLC1 tumor suppressor, mRNA, was shown to be a target for miR-216a. It was also reported that the reduction of TSLC1 (CADM1) expression correlated with the up-regulation of miR-216a.[16] When the expression of CADM1 was determined in an HCC gene-expression database reported on previously by us[9] and in the Instructional Systems Technology (IST) online system (a repository of genomics database; http://www.medisapiens.com/ist-online-overview/), it was demonstrated that the expression of CADM1 was up-regulated in HCC patients with early recurrent disease (Supporting Figs. 4D and 9A). These results suggest an alternate regulatory mechanism, instead of through the androgen-signaling pathway, for regulating the expression of the miR-216a/217 cluster, contributing to the recurrence and metastatic potential of HCC.

PTEN and SMAD7 have been identified as two putative targets for miR-216a/217 using several different miRNA target-prediction programs and experimental validation.[16] Expression of PTEN and SMAD7 was significantly down-regulated in HCC samples, compared to matched adjacent normal and histologically normal liver tissue. Furthermore, PTEN and SMAD7 were strongly down-regulated in samples from patients with early recurrence (Fig. 4C and 4D) and were significantly associated with the DFS of these patients (Fig. 4E,F). Although there are many reports of the inactivation of PTEN in cancers and its association with the advanced stages of cancers and metastases, the molecular mechanism of PTEN in HCC tumor recurrence and metastases is not well characterized. An earlier study has shown that PTEN was underexpressed in HCC, compared to corresponding nontumorous liver tissue, and the underexpression of PTEN was mediated by the AKT/SP1/matrix metalloproteinase 2–signaling pathway and associated with poorer patient survival.[23] More recently, Wu et al. studied the mechanism underlying the progression from cirrhosis to HCC and showed that the level of TGF-β was positively correlated with hepatic tumor-initiating cells.[24] Moreover, hyperactivation of Akt, but not Notch, signal transducer and activator of transcription 3, or mammalian target of rapamycin, was detected in rat pluripotent liver progenitor cell-like WB-F344 cells treated with TGF-β. Furthermore, TGF-β-induced Akt activation and liver progenitor cell transformation was mediated by miR-216a-modulated PTEN suppression.[24] Our experiments with human HCC cells corroborated well with these data, and we have further demonstrated that the overexpression of miR-216a/217 can act as a positive feedback regulator for the TGF-β pathway in HCC through the novel target, SMAD7. SMAD7 was shown to be down-regulated in our HCC gene-expression database[9] and in the IST online system (Supporting Figs. 4A and 9B). SMAD7 is a member of the SMAD family of proteins, which belong to the TGF-β superfamily of ligands. SMAD7 is involved in cell signaling and is a TGFBR1 antagonist that blocks TGFB1.[18, 20] In this context, the miR-106b-25 cluster has been shown to activate TGF-β signaling by targeting SMAD7 to induce EMT and tumor-initiating cell characteristics in human breast cancer.[25]

In conclusion, we identified that up-regulation of the miR-216a/217 cluster was associated with early recurrence, survival, and EMT phenotype in HCC tissue and cell lines. Overexpression of miR-216a/217 induced EMT and cancer stem-like properties by activating the TGF-β- and PI3K/Akt-signaling pathways through down-regulation of PTEN and SMAD7. Additionally, activation of the TGF-β- and PI3K/Akt-signaling pathways in HCC cells resulted in an acquired resistance to sorafenib, whereas blocking activation of the TGF-β pathway overcame miR-216a/217-induced sorafenib resistance and prevented tumor metastases (Fig. 7). These results may provide novel prognostic and predictive factors for early-recurrent HCC disease and the design of novel miRNA-based therapeutic strategies against HCC.

Figure 7.

Schematic diagram depicting possible regulatory roles of miR-216a/217 contributing to EMT, metastases, recurrence, and sorafenib resistance in HCC (→ = activation; ┤= inhibition; red indicates up-regulation and green indicates down-regulation).

Acknowledgment

Immunohistochemical staining was performed by the Advanced Molecular Pathology Laboratory, The Institute of Molecular and Cell Biology, Singapore's Agency for Science, Technology and Research.

Ancillary