MicroRNA-122 Triggers Mesenchymal-Epithelial Transition and Suppresses Hepatocellular Carcinoma Cell Motility and Invasion by Targeting RhoA

The loss of microRNA-122 (miR-122) expression is strongly associated with increased invasion and metastasis, and poor prognosis of hepatocellular carcinoma (HCC), however, the underlying mechanisms remain poorly understood. In the present study, we observed that miR-122 over-expression in HCC cell lines Sk-hep-1 and Bel-7402 triggered the mesenchymal-epithelial transition (MET), as demonstrated by epithelial-like morphological changes, up-regulated epithelial proteins (E-cadherin, ZO-1, α-catenin, occludin, BVES, and MST4), and down-regulated mesenchymal proteins (vimentin and fibronectin). The over-expression of miRNA-122 also caused cytoskeleton disruption, RhoA/Rock pathway inactivation, enhanced cell adhesion, and suppression of migration and invasion of Sk-hep-1 and Bel-7402 cells, whereas, these effects could be reversed through miR-122 inhibition. Additional studies demonstrated that the inhibition of wild-type RhoA function induced MET and inhibited cell migration and invasion, while RhoA over-expression reversed miR-122-induced MET and inhibition of migration and invasion of HCC cells, suggesting that miR-122 induced MET and suppressed the migration and invasion of HCC cells by targeting RhoA. Moreover, our results demonstrated that HNF4α up-regulated its target gene miR-122 that subsequently induced MET and inhibited cell migration and invasion, whereas miR-122 inhibition reversed these HNF4α-induced phenotypes. These results revealed functional and mechanistic links among the tumor suppressors HNF4α, miR-122, and RhoA in EMT and invasive and metastatic phenotypes of HCC. Taken together, our study provides the first evidence that the HNF4α/miR-122/RhoA axis negatively regulates EMT and the migration and invasion of HCC cells.


Introduction
Hepatocellular carcinoma (HCC) is one of the most prevalent human malignancies. The elucidation of the molecular mechanisms underlying the tumorigenicity, invasion and metastasis of HCC is critically important for the development of novel treatments for this disease.
The epithelial-mesenchymal transition (EMT), a reverse process of MET, is a critical event in tumor invasion and metastasis that cause the majority of cancer death [29][30][31]. Therefore, understanding the molecular mechanisms of EMT may lead to the development of novel interventions for cancer metastasis.
Given the important roles of miR-122 in HCC invasion and metastasis [16,17,21,23,25,27,28], we aim to investigate the mechanisms of miR-122-induced MET and inhibition of migration and invasion. The candidate transcription factor regulating the expression of miR-122 and the molecular target of miR-122 were also identified and investigated.

Cell lines and cell culture
The human HCC cell lines, Bel-7402 and Sk-hep-1, were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). HEK293T cells were obtained from the American Type Culture Collection (ATCC). The cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) in a humidified incubator with 5% CO 2 at 37uC.

RNA isolation and quantitative real-time PCR
For mRNA analyses, total RNA from HCC cells was extracted using Trizol Reagent (TaKaRa) according to the protocol provided by the manufacturer. Total RNA was reversely transcribed with the PrimeScript RT reagent Kit (TaKaRa). Quantitative real-time PCR (qRT-PCR) for the expression of mRNA analysis was performed using SYBR Green qRT-PCR master mix (TaKaRa) as described using GAPDH for normalization on a Stratagene Mx3005P qRT-PCR System. The primers used for the amplification of the indicated genes were listed in Table S1 and S2 in File S1. All samples were normalized to internal controls and fold changes were calculated through relative quantification (2 2ggCt ).

Western blot analysis and Rho-GTPase activation assay
Protein lysates were separated by SDS-PAGE, and electrophoretically transferred to PVDF (polyvinylidene difluoride) membrane. Then, where relevant, the blots were probed with the primary antibodies as labeled in the corresponding Figs, followed by HRP (horseradish peroxidase)-labeled goat anti-mouse or goat anti-rabbit IgG, and the signals were detected using enhanced chemiluminescence (ECL). b-actin was used as a protein loading control. For origin and description of all antibodies used in this study see Table S3 in File S1.

F-actin cytoskeleton staining analysis
F-actin staining was performed according to the manufacturer's instructions. Fixed cells were incubated with 100 nM working stock of Acti-stain 488 phalloidin (Cytoskeleton). Cells were counterstained with DAPI (Sigma) and imaged with a confocal laser-scanning microscope (Olympus FV1000).

Transwell migration assay and Boyden invasion assay
For transwell migration assay, 1610 5 cells were seeded into the upper chamber (BD Biosciences, MA) with serum-free DMEM. Boyden invasion assay was performed using matrigel (BD Biosciences, MA) in the upper chamber. DMEM with 10% FBS were put into lower compartment as chemo-attractant. Cells were allowed to migrate for 12 hours. Remaining cells in the upper chamber were scraped out by cotton swap. The cells that had migrated to the lower surface of the membrane were fixed with 100% methanol and stained with hematoxylin solution (Sigma), followed by counting in five random optical fields.

Cell adhesion assay
Cell adhesion experiment was carried out according to the methods described previously [34]. In brief, cells were seeded on culture dishes and grown to ,90% confluence. The confluent monolayers were then washed with PBS and treated with 0.05 mM EDTA. Cells detachment was examined at different time intervals (2, 10, 20 and 30 min) under a phasecontrast microscope. Cell adhesion was analyzed by counting the adherent cells.

mRNA microarray analysis
Expression microarray analysis was carried out with commercially available Affymetrix Human Gene U133 Plus 2.0 array according to the Affymetrix standard protocol, which carried 47,000 transcripts representing 38,500 well-characterized human genes. All the hybridization procedures and data analysis were performed by Capital Bio Corp. (Bejing, China). Total RNA samples were isolated from liver cancer cells (Sk-hep-1 cells) using Trizol reagent (Invitrogen). Briefly, total RNA was used to synthesize cDNA in an in vitro transcription reaction, and then cDNA was fluorescently labeled by Cy5 or Cy3-CPT with Klenow enzyme. Labeled cDNA was then hybridized to Affymetrix Human Gene U133 Plus 2.0 arrays. Hybridization was processed at 45uC, with rotation for 16 h (Affymetrix GeneChip Hybridization Oven 640). Chips were then washed and stained in the Affymetrix Fluidics Station 450. Hybridization signals were scanned with a Lux-Scan 3.0 scanner (Capital Bio. Corporation, Beijing, China). The resultant images were digitized with Genepix Pro 6.0 software (Axon Instruments, Foster City, CA, USA).

Scanning electron microscopy
Scanning electron microscopy was performed according to the methods described previously [34]. Cells were fixed with 1% osmium tetroxide and 2.5% glutaldehyde followed by stepwise ethanol dehydration. After the step of critical point drying, slides were mounted on silver paste. Images were scanned and captured under 63,000 and 66,000 magnifications using the quanta 200 scanning electron microscope (Nikon S-3000N). Data were processed with Adobe Photoshop 7.0 software.

Statistical analysis
Data were presented as mean6SD unless otherwise indicated of at least 3 independent experiments. Statistical analysis was performed using a SPSS 13.0 software package. Two-tailed Student's t test was used for comparisons of 2 independent groups. Statistical significance was assessed by the Student's t-test (*p,0.05; **p,0.01).

miR-122 trigged MET and inhibited the motility and invasion of HCC cells
The miR-122 transgene was over-expressed in Sk-hep-1 and Bel-7402 cells ( Figure S1 in File S1), and these cells underwent significant epithelial-like morphological changes ( Figure 1A), suggesting that miR-122 induced MET. Therefore, we characterized the MET at the molecular level by measuring the expression levels of a number of proteins markers. The results of the qRT-PCR analysis showed increased expression of epithelial markers, such as cell adhesion proteins (E-cadherin and BVES) and tight junction protein (occludin), and reduced expression of mesenchymal markers (vimentin and fibronectin) in miR-122-expressing Skhep-1 and Bel-7402 cells ( Figure 1B Table S4 in File S1). A miR-122 inhibitor was transiently transfected into miR-122expressing Bel-7402 cells to confirm that miR-122 induced MET. We found that the inhibition of miR-122 reversed the MET phenotypes observed in miR-122-expressing cells ( Figures 1A, B, C).
The loss of cell adhesion, inhibition of E-cadherin expression, and increase in cell motility are characteristics of EMT, the reverse process of MET [34]. As shown in Figure 2A, miR-122-expressing Sk-hep-1 and Bel-7402 cells exhibited epithelial-like phenotypes with significantly reduced cell migration and invasion and significantly enhanced cell adherence ( Figure 2B). The inhibition of miR-122 reversed these epithelial-like phenotypes induced by miR-122 over-expression.
The ectopic expression of miR-122 inactivated the RhoA/ Rock pathway and impaired cytoskeletal events The RhoA/Rock pathway is involved in EMT and MET by regulating cytoskeletal events such as the formation of stress fiber bundling arrays [29][30][31]. Therefore, we characterized the status of stress fiber formation and actin polymerization in miR-122expressing cells. Using phalloidin staining, we observed that stress fiber formation was suppressed in miR-122-expressing cells, but not in empty vector-expressing cells. In addition, cytoskeletal disruption in miR-122-expressing cells could be restored by a miR-122 inhibitor ( Figure 3A).
Filopodia and lamellipodia, dynamic structures on cell membrane, are involved in cell motility and cancer cell invasion and metastasis. The formation of filopodia and lamellipodia requires actin polymerization. Under a scanning electron microscope, a significant reduction or loss of filopodia and lamellipodia (cell protrusion) was observed on the membrane of miR-122-expressing cells, whereas miR-122 inhibition restored the formation of filopodia and lamellipodia in miR-122-expressing Bel-7402 cells ( Figure 3B).
Previous studies have reported that RhoA induced the formation of stress fibers and Rac1 and Cdc42 stimulated the formation of lamellipodium and filopodia [29][30][31]. Pull-down assays were employed to investigate the active forms of Rho-GTPases. As shown in Figure 3C, higher levels of the active forms of RhoA, Rac1 and Cdc42 were observed in empty vector-expressing cells compared with miR-122-expressing cells. In addition, the RhoA/Rock pathway could be reactivated by the inhibition of miR-122 in Bel-7402 cells.

miR-122 induced MET and inhibited the migration and invasion of HCC cells by targeting RhoA
First, RhoA and Rac1 were identified as two potential targets of miR-122 by searching the miRNA databases microRNA.org, RNAhybrid, and miRWalk. The 39-UTR of RhoA and Rac1 mRNAs contains a complementary site for the seed region of miR-122 ( Figure 4A). Notably, previous studies have shown that RhoA and Rac1 played central roles in cell motility by regulating the cytoskeletal signaling pathway [29][30][31].
To determine whether miR-122 regulates RhoA and Rac1, we measured changes in RhoA and Rac1 expressions in HCC cells after the over-expression or inhibition of miR-122. The results of Western blotting revealed that total and active forms of both RhoA and Rac1 were dramatically reduced after the ectopic expression of miR-122 in Sk-hep-1 and Bel-7402 cells ( Figure 3C). In addition, the knockdown of miR-122 using a miR-122 inhibitor increased the levels of total and active forms of RhoA and Rac1 ( Figure 3C). Therefore, miR-122 negatively regulates RhoA and Rac1 in HCC cells.
We performed a luciferase reporter assay to determine whether miR-122 directly targets the 39-UTRs of RhoA and Rac1. Transient transfection with wild-type RhoA-luc reporter and miR-122 mimics into 293T and Bel-7402 cells significantly reduced luciferase activity compared with mimics control ( Figure 4A), whereas transient transfection with wild-type Rac1luc reporter and miR-122 mimics could not significantly reduce the luciferase activity ( Figure 4A). The results of qRT-PCR showed that RhoA mRNA levels were reduced in miR-122expressing Sk-hep-1 and Bel-7402 cells, and the knockdown of miR-122 using a miR-122 inhibitor restored the levels of RhoA mRNA ( Figure S2 in File S1). Therefore, these results suggest that RhoA, rather than Rac1, is a direct target of miR-122.
To elucidate whether the miR-122-induced MET and inhibition of cell motility and invasion are mediated through RhoA in HCC cells, we performed gain-and loss-of-function experiments. First we inhibited the function of wild-type RhoA using a  Figure 4C).
Subsequently, we evaluated whether the ectopic expression of RhoA reverses the miR-122-induced MET and rescues the miR-122-mediated suppression of migration and invasion of Bel-7402 cells. As expected, RhoA over-expression reversed the MET phenotypes ( Figure 4B) and abrogated the suppressed motility and invasion ( Figure 4C) induced by miR-122 in Bel-7402 cells.
Taken together, these results suggest that RhoA is the target of miR-122 in the regulation of EMT, motility and invasion of HCC cells.

miR-122 was regulated by the transcriptional factor HNF4a to induce MET and inhibit the motility and invasion of HCC cells
Previous studies have demonstrated that miR-122 was positively associated with the expression of the liver-enriched transcription factors, such as HNF1A, HNF3A, HNF3B, HNF4A (HNF4a), HNF4G, and HNF6 [17], essential for hepatocyte differentiation.
For example, it has been reported that HNF4a bound to the miR-122 promoter region to regulate miR-122 expression in HCC cells in mouse livers [35]. Other studies have also reported that HNF4a inhibited cell proliferation, EMT, invasion and metastasis of HCC cells [17,[36][37][38][39]. Therefore, we aimed to determine whether HNF4a directly regulates miR-122-induced MET and inhibition of migration and invasion of HCC cells.
We found that the ectopic expression of HNF4a ( Figure S3 in File S1) increased miR-122 expression in Bel-7402 cells ( Figure S4 in File S1), and HNF4a-expressing cells underwent significant epithelial-like morphological changes ( Figure 5A). Western blot analysis validated the increase in ZO-1 and a-catenin expressions and the reduction of vimentin and fibronectin expressions in HNF4a-expressing Bel-7402 cells ( Figure 5B). Therefore, HNF4a over-expression induced epithelial-like phenotypes and MET-like cellular marker alterations in HCC cells, similar to the results of restoring miR-122 expression in HCC cells ( Figure 1C).
We also studied the changes in the motility and invasion of HNF4a-expressing cells. Similar to miR-122 over-expression (Figure 2A), HNF4a-expressing Bel-7402 cells exhibited significantly decreased mobility and invasion ( Figure 5C). In addition, the repression of endogenous miR-122 expression by a miR-122 inhibitor reversed HNF4a-induced MET-like cellular marker alterations ( Figure 5B) and abolished HNF4a-suppressed motility and invasion of Bel-7402 cells ( Figure 5C). Taken together, these results suggest that HNF4a is involved in the EMT, motility and invasion of HCC cells by regulating miR-122.

Discussion
The majority of cancer deaths reflect tumor metastasis, thus the prevention of metastasis has received much attention [29][30][31]. EMT is an early event that is critical for cancer invasion and metastasis. To facilitate cell motility and invasion, epithelial cells lose adherence, tight junction proteins, polarity and cell-cell contact and undergo remarkable cytoskeleton remodeling [29][30][31]. Thus, understanding the molecular mechanisms of EMT will help elucidate the events involved in the metastatic cascade and might lead to the development of novel therapeutics for treating metastatic cancers. Previous studies have shown that miR-122 negatively regulated invasion and metastasis in HCC [16,17,21,23,25,27,28]. However, the underlying mechanisms are not well understood. The results of the present study demonstrated that the reintroduction of miR-122 into Sk-hep-1 and Bel-7402 cells induced epithelial-like morphological conversion, accompanied by the increased expression of epithelial markers (E-cadherin, ZO-1 and a-catenin), decreased expression of mesenchymal markers (vimentin and fibronectin), increased cell adhesion, cytoskeleton depolymerization, RhoA/ Rock pathway inactivation, and decreased cell motility and invasion. In addition, the knockdown of miR-122 reversed these phenotypes. These results are consistent with previous studies, showing that epithelial-like phenotypes resulted from the overexpression of miR-122 in hepG2 [16,25] and Malhlavu cells [28], accompanied by the up-regulation of E-cadherin [25] and down-regulation of vimentin [16,28]. Furthermore, the DNA microarray results highlighted the corresponding molecular features of miR-122-expressing Sk-hep-1 cells (Figures 1D, E and Table S4 in File S1), strongly supporting the aforementioned biological transformations. Collectively, the results of the present study and those obtained in other studies [16,25,28] fully support that the ectopic expression of miR-122 in liver cancer cells triggers MET that may contribute to the inhibition of motility and invasion of cancer cells.
As a cell moves in a designated direction, the cooperation between continuous actin polymerization and depolymerization facilitates the protrusion of the cell at the anterior front end [40]. In addition, the cell undergoes consecutive actomyosin contractions and separates from the posterior end, resulting in directional cell movement [40]. Moreover, the formation of the stress fibers, filopodia and lamellipodia also requires actin polymerization [40]. The loss of these structural features indicates an inability to engage in actin polymerization, resulting in reduced cell motility and invasion [40]. The results of the present study demonstrated that the exogenous expression of miR-122 in Sk-hep-1 and Bel-7402 cells substantially inhibited the RhoA/Rock pathway and suppressed the formation of stress fibers, filopodia and lamellipodia, whereas the inhibition of miR-122 restored the formation of stress fibers, filopodia and lamellipodia. These observations suggest that miR-122 suppressed HCC cell invasion and metastasis by regulating the formation of stress fibers, filopodia and lamellipodia.
Notably, RhoA is a key component of the RhoA/Rock pathway, a well-characterized regulator of cytoskeletal organization and a positive regulator of EMT and the motility, invasion and metastasis of cancer cells, including HCC cells [29][30][31]. Thus, miR-122 induces MET and negatively modulates cell motility and invasion through the direct inhibition of RhoA in HCC cells.
In the present study, we observed that loss of RhoA function through RhoA-T19N largely mimicked miR-122-induced MET, and the inhibition of migration and invasion of tumor cells, whereas RhoA over-expression in miR-122-expressing Bel-7402 cells reversed miR-122-induced MET and abrogated the miR-122-mediated suppression of motility and invasion of HCC cells. These findings suggest that miR-122 induced MET and inhibited the migration and invasion of HCC cells by targeting RhoA. To play multiple roles in diverse biological processes, miRNAs are mainly regulated by upstream transcriptional factors [50]. Previous studies have demonstrated that miR-122 was strongly associated with HNF4a expression in HCC tissues. Moreover, it has been reported that miR-122 is a direct target of HNF4a [35]. Similar to miR-122, the loss of HNF4a expression in HCC has been associated with a gain of invasive and metastatic properties [17]. However, whether HNF4a directly targets miR-122 to regulate miR-122-induced MET and inhibition of migration and invasion of HCC cells has not yet been reported. The results of the present study demonstrated that the ectopic expression of HNF4a induced MET to inhibit HCC cell motility and invasion through the up-regulation of miR-122 expression. Both HNF4a [17,36,37] and miR-122 [16,17,21,23,25,27,28] act as tumor suppressors and negative regulators of HCC invasion and metastasis, whereas RhoA positively regulates HCC invasion and metastasis [41,43,49]. Furthermore, miR-122 overexpression could promote hepatic differentiation and maturation of mouse embryonic stem cells (ESC)-derived cells by regulating the balance between EMT and MET at least in part through a miR-122/FoxA1/HNF-4a positive feedback loop [51]. Based on the results of the present study and previous studies [16,17,21,23,25,27,28,[35][36][37]41,43,49], we propose that HNF4a induces miR-122 expression, which subsequently suppresses RhoA, thereby inactivating the RhoA/ Rock pathway, enhancing cell adhesion and cell junction, impairing cytoskeletal events, and decreasing cell motility ( Figure 6). In summary, these results illustrate that the HNF4a/ miR-122/RhoA axis negatively regulates EMT and the invasion and metastasis of HCC cells ( Figure 6).
In conclusion, these findings demonstrate, for the first time, that miR-122 that is regulated by the upstream transcriptional factor HNF4a triggers MET to suppress HCC cell motility and invasion by targeting RhoA. Thus, miR-122 might be a promising therapeutic target for the inhibition of invasion and metastasis of HCC.