miR-151-3p Targets TWIST1 to Repress Migration of Human Breast Cancer Cells

TWIST1 is a highly conserved basic helix-loop-helix transcription factor that contributes to cancer metastasis by promoting an epithelial-mesenchymal transition and repressing E-cadherin gene expression in breast cancer. In this study, we explored the potential role of miR-151 in TWIST1 expression and cancer properties in human breast cancer cells. We found that the human TWIST1 3’UTR contains a potential binging site for miR-151-3p at the putative target sequence 5’-CAGUCUAG-3’. Using a TWIST1-3’UTR luciferase reporter assay, we demonstrated that the target sequence within the TWIST1 3’UTR is required for miR-151-3p regulation of TWIST1 expression. Moreover, we found that ectopic expression of miR-151-3p by infection with adenoviruses expressing miR-151 significantly decreased TWIST1 expression, migration and invasion, but did not affect cell growth and tumorsphere formation of human breast cancer cells. In addition, overexpression of the protein coding region without the 3’UTR of TWIST1 reversed the repression of cell migration by miR-151-3p. Furthermore, knockdown of miR-151-3p increased TWIST1 expression, reduced E-cadherin expression, and enhanced cell migration. In conclusion, these results suggest that miR-151-3p directly regulates TWIST1 expression by targeting the TWIST1 3’UTR and thus repressing the migration and invasion of human breast cancer cells by enhancing E-cadherin expression. Our findings add to accumulating evidence that microRNAs are involved in breast cancer progression by modulating TWIST1 expression.


Introduction
Breast cancer is one of the most common malignancies in women and its incidence rate is increasing [1]. Although there has been a remarkable improvement in mortality from breast cancer, recurrence and metastases remain the major causes of death for breast cancer patients [2]. Understanding the mechanisms responsible for breast cancer progression and developing more specifically targeted, less toxic therapies are critical issues in breast cancer treatment.
In human breast cancers, TWIST1 is usually found to be over-expressed, which is correlated with invasive lobular carcinoma, a highly infiltrating tumor type associated with loss of E-cadherin expression, lymph-node and distant metastases, and poor patient prognosis [3][4][5][6]. TWIST1 is a highly conserved basic helix-loop-helix (bHLH) transcription factor and is characterized by a basic DNA binding domain that targets the consensus E-box sequence 5'-CANNTG-3' and a helix-loop-helix domain [7]. TWIST1 contributes to cancer metastasis by promoting an epithelial-mesenchymal transition (EMT) [8,9]. Moreover, TWIST1 is a transcriptional repressor of E-cadherin gene expression in breast cancer [10]. Based on the function of E-cadherin as a cell-cell adhesion molecule, loss of E-cadherin is considered a prerequisite for EMT favoring tumor cell dissemination and metastasis [8]. Therefore, the regulation of TWIST1 expression in cancer cells might be a potential target for the suppression of cancer cell metastases.
MicroRNAs (miRNAs) are endogenous small single-stranded non-coding RNAs, typically 20-22 nucleotides in length, that regulate gene expression by binding specific sequences in the 3'-untranslated region (3'-UTR) of the target mRNA [11,12]. Accumulating evidence has proven that deregulation of miRNA is involved in a wide range of human diseases, including cancer [13]. In human cancer, miRNAs can function as oncogenes or tumor suppressor genes during tumorigenesis, depending on their target genes [14]. Recently, some miRNAs were identified to modulate cancer properties by directly targeting TWIST1 expression in different cancer cells [15], suggesting that TWIST1 might be regulated by different miRNAs during cancer progression.
In this study, we adopted in silico analyses and found that the TWIST1 3'UTR contains a potential binging site for miRNA (miR)-151-3p at the putative target sequence from nucleotide position (np) 71 to np 87. The miR-151 gene localizes to chromosome 8q24.3 and resides within intron 22 of the host gene encoding focal adhesion kinase (FAK) [16]. It has been reported that miR-151 regulates tumor cell migration and spreading of hepatocellular carcinoma (HCC) [16,17]. Downregulating Rho GDP Dissociation Inhibitor (GDI) Alpha (RhoG-DIA) by miR-151 enhanced HCC cell migration through the activation of Rac1, Cdc42 and Rho GTPases [16]. In breast cancer, miR-151-5p expression levels were not different among tumors of varying grades, but the level was significantly lower in the lymph-node metastases than in their corresponding tumors of breast cancer patients [18]. It was recently demonstrated that miR-151-5p combined with other miRNAs (miR-145a-5p or miR-337-3p) are able to significantly repress TWIST1 translation and result in the decreased migratory potential of murine embryonic fibroblast cells [19]. However, the role of miR-151 in breast cancer progression and its direct targets in the regulation of breast cancer metastasis are still undefined. In this study, we explored the potential role of miR-151 in TWIST1 expression and cancer properties in human breast cancer cells.

Recombinant adenoviruses
The recombinant miR-151-expressing adenoviral plasmid was linearized by PacI and subsequently transfected into the adenovirus packaging cells AD-293 to obtain packaged recombinant adenovirus expressing miR-151 (designated Ad-miR-151). The pAdTrack-CMV empty vector was used to generate recombinant adenovirus expressing GFP as a control (designated Ad-GFP). The titer and multiplicity of infection of recombinant adenoviruses were determined according to the manufacturer's protocol (Stratagene).

Quantitative real-time PCR analysis
Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA), and cDNA was synthesized using Moloney murine leukemia virus reverse transcriptase (New England BioLabs, Beverly, MA) with an oligo (dT 18 ) primer. Quantitative real-time PCR was performed using a Light Cycler system with a LightCycler FastStart DNA Master PLUS SYBR Green I (Roche, Indianapolis, IN) [21]. The primers used in real-time PCR analysis are TWIST1-f: The relative mRNA levels were normalized to that of GAPDH. For the detection of mature miRNAs, cDNA synthesis and TaqMan miRNA real-time PCR assays were performed as previously described [22,23]. The relative miRNA levels were normalized to the level of RNU48, which is a small nucleolar RNA.

Cell proliferation assay
Cell proliferation was determined using the MTT assay. The cells (3 × 10 4 ) were seeded in 24-well plates cultured overnight prior to infection. After infection with adenoviruses expressing miR-151 or GFP for 48 hours, the medium was discarded and replaced with 0.25 mg/ml 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide (MTT; Sigma-Aldrich) and incubated for 2 hours at 37˚C in the dark. The reagent was discarded, and the plates were washed with PBS. The cells were combined with 250 μl dimethyl sulfoxide (DMSO) to dissolve the formazan produced. The absorbance at 550 nm was measured using a microplate enzymelinked immunosorbent assay reader (TECAN Infinite 200). We further determined cell growth using the sulforhodamine B (SRB) assay. After infection with adenoviruses expressing miR-151 or GFP for 48 hours, MDA-MB-231 cells (5 × 10 3 ) were seeded in 96-well plates for 0, 24, 48, 72 and 96 hours. The cells were fixed with 10% ice-cold trichloroacetic acid (TCA) (Sigma- Aldrich) at 4˚C for 1 hour, rinsed four times with distilled water and air dried. The cells were then stained with 0.057% SRB (Sigma-Aldrich) in 1% acetic acid for 30 min at room temperature. After rinsing four times with 1% acetic acid and air drying, 50 μl of 10 mM Tris-base (pH 10.5) was added to each well for 30 min. The colorimetric level was read using a microplate reader (Tecan) at 510 nm.

Migration and invasion assays
Cellular migration and invasion abilities (1 × 10 4 for migration assay, 5 × 10 4 for invasion assay) were analyzed in 24-well plates using Millicell tissue culture plate well inserts (Millipore, Bedord, MA) for 12 hours and BD BioCoat Matrigel Invasion Chambers (Becton Dickson, Mountain View, CA) for 20 hours, respectively. A total of 5 × 10 4 breast cancer cells were placed in the serum-free medium inside the inserts and the 10% FBS medium was added in the lower chamber. Then, the cells on the upper surface of membrane were removed with a cotton swab after incubation. After fixation with methanol, cells in the lower surface of membrane were stained with 0.005% crystal violet in PBS for 1 hour, and the number of migrated or invaded cells was determined by counting 10 random fields under a microscope.

Statistical analysis
All data are presented as the mean ± SEM (standard error of the mean) from at least 3 independent experiments. Significant differences between the control and treatment groups were determined using Student's t-test, and the significance was considered at P < 0.05.
To validate whether the identified binding site is important for miR-151-3p regulation of TWIST1 expression, we constructed two luciferase reporter plasmids, the wild type TWIST1 3'UTR (TWIST1-3'UTR-WT) and the 8 bp-deletion mutant (TWIST1-3'UTR-8bp-deletion). The two reporter plasmids were successfully constructed and confirmed by DNA sequencing (Fig 1B). Using the two luciferase reporter constructs, we observed that the ectopic expression of miR-151 (Ad-miR-151) can significantly decrease the reporter signals of the TWIST1-  3'UTR-WT, but not significantly decrease that of the TWIST1-3'UTR-8bp-deletion when compared with the control (Ad-GFP) groups in HEK293T (Fig 1C) and MDA-MB-231 cells (Fig 1D). These results indicate that the sequence 5'-CAGUCUAG-3' within the TWIST1 3'UTR is required for miR-151-3p regulation of TWIST1 expression in human cells.

miR-151 expression is increased in human breast cancers
To evaluate the miR-151 expression in breast cancers, we analyzed the data obtained from The Cancer Genome Atlas (TCGA) and found that miR-151 expression levels and its host gene FAK mRNA levels in breast cancers were significantly higher when compared with that of paired normal tissues (Fig 2A). Moreover, the miR-151 expression levels were significantly positively correlated with the FAK mRNA levels in breast cancers (Fig 2B). Although no significant difference was observed among the stages of breast cancer, the miR-151 and FAK expression levels of breast cancers at different stages were higher than those of normal tissues ( Fig  2C). These results suggest that the miR-151 expression is increased in breast cancer and is associated with the expression of its host gene FAK.
In addition, after infection with adenoviruses expressing miR-151, we found that the mRNA ( Fig 3A) and protein ( Fig 3B) expression levels of TWIST1 were significantly decreased in human breast cancer cells. The results further confirmed the role of miR-151-3p in regulating TWIST1 expression in human breast cancer cells. Interestingly, it was noted that the data obtained from TCGA showed that there is a significantly negative correlation between the miR-151 expression levels and TWIST1 mRNA levels ( Fig 3C). We further evaluated that the correlations between miR-151 and TWIST1-related gene expression levels in human breast cancers and found that miR-151 levels is significantly positive correlated with E-cadherin gene expression levels, and is significantly negative correlated with the gene expressions of fibronectin, osteonectin, and α5-integrin (Table 1). Among these TWIST1-related genes, the expression E-cadherin is repressed by TWIST1, while the expressions of fibronectin, osteonectin,  and α5-integrin are activated by TWIST1 [25]. These results suggest that miR-151-3p might regulate TWIST1 expression and thereby TWIST1-related downstream gene expression.

miR-151-3p represses cell migration and invasion, but not cell growth and tumorsphere formation of human breast cancer cells
To evaluate the biological function of miR-151-3p in human breast cancer cells, we ectopically expressed miR-151 in three breast cancer cell lines (MCF-7, MDA-MD-231, and HCC1937) by infection with adenoviruses expressing miR-151 and analyzed the effects on cell growth, tumorsphere formation, cell migration and invasion. The results of the MTT assay ( Fig 4A) and SRB assay ( Fig 4B) revealed that the ectopic expression of miR-151 does not significantly alter cell growth when compared with that of controls. In addition, the number and size of the tumorspheres within HCC1937 cells were not significantly affected by ectopic expression of miR-151 (Fig 4C). Using a transwell migration assay and a Matrigel invasion assay, we found that both migration ( Fig 4D) and invasion (Fig 4E) of the three human breast cancer cell lines were significantly repressed by infection with adenoviruses expressing miR-151. These results indicate that miR-151-3p can regulate cell migration and invasion, but not cell growth and tumorsphere formation in human breast cancer cells.

Overexpression of the TWIST1 protein-coding region reverses miR-151-3p inhibition of cell migration
To validate the role of TWIST1 expression in the miR-151-3p regulation of cell migration, we transfected MDA-MB-231 cells with the TWIST1 cDNA plasmid to overexpress the protein coding region (that lacks the TWIST1 3'UTR) after infection with adenoviruses expressing miR-151. We found that overexpression of TWIST1 can reverse the TWIST1 expression that is decreased by ectopic expression of miR-151 ( Fig 5A). Importantly, overexpression of TWIST1 can overcome the miR-151's inhibitory effect on the migration of MDA-MB-231 cells (Fig 5B). These results suggest that miR-151-3p may regulate cell migration through targeting TWIST1 expression.

Knockdown of miR-151-3p increases TWIST1 expression and enhances cell migration of human breast cancer cells
TWIST1 is known as a critical regulator of EMT and is a transcriptional repressor of Ecadherin gene expression in breast cancer. We analyzed the expression of EMT-related genes in MDA-MD-231 cells after infection with adenoviruses expressing miR-151 (Ad-miR-151) and found that the gene expression of E-cadherin was significantly increased, whereas the gene expression levels of β-catenin, slug, N-cadherin, snail and vimentin were not significantly different when compared with that of the control (Ad-GFP) (Fig 6A). The results suggest that E-cadherin may be involved in miR-151-3p-TWIST1-mediated cell migration.   In addition, we used antogomir-151-3p (anti-miR-151-3p) to knockdown endogenous miR-151-3p expression in breast cancer cells. Fig 6B shows that the miR-151-3p expression was significantly decreased by anti-miR-151-3p in a dose-dependent manner. Moreover, knockdown of miR-151-3p increased TWIST1 protein expression and decreased E-cadherin protein expression (Fig 6C). In addition, the migration of HCC1937 cells was significantly increased by anti-miR-151-3p treatment when compared with that of the scrambled control (Fig 6D). These results support the notion that miR-151-3p regulates TWIST1 expression and thus modulates E-cadherin expression as well as cell migration of breast cancer cells.

Discussion
This is the first study to demonstrate that miR-151-3p directly regulates TWIST1 expression by targeting the TWIST1 3'UTR and thus represses the migration and invasion of human breast cancer cells by enhancing E-cadherin expression. Our findings add to accumulating evidence that microRNAs are involved in breast cancer progression by modulating TWIST1 expression.
Several miRNAs have been documented to directly inhibit TWIST1, including miR-1-1 [26], miR-33a [27,28], miR-137 [29], miR-186 [30], miR-300 [31], miR-520d-5p [32], miR-539 [31], miR-543 [31], miR-675 [33], and miR-720 [34]. Most of these miRNAs repressed cancer cell EMT and metastasis in most examined cancer types, and some of them (miR-33a and miR-186) modulated cancer cell sensitivity to cisplatin by down-regulating TWIST1 [27,30]. For example, it was demonstrated that miR-33a is upregulated in chemoresistant osteosarcoma and promotes cell resistance to cisplatin [27]. On the other hand, miR-186 was shown to induce G1 cell-cycle arrest and enhance cell apoptosis, which consequently rendered the cells more sensitive to cisplatin in ovarian cancers [30]. Among these miRNAs, the data obtained from TCGA reveal that like miR-151, miR33a and miR-137 are significantly higher expression levels in breast cancers as compared with the paired normal tissues (Fig 7). The expression levels of miR-520d, miR-539, miR-543, and miR-675 in breast cancers were significantly lower than paired normal tissues (Fig 7). Interestingly, we found that miR-33a and miR-151 are significantly negative correlated with TWIST1 gene expression in breast cancers (Table 2). In the present study, we found that ectopic expression of miR-151 represses cell migration and invasion of breast cancer cells by down-regulating TWIST1 (Figs 4D, 4E and 5B). In addition, knockdown of miR-151-3p by anti-miR-151-3p can up-regulate TWIST1 expression and enhance cell migration (Fig 6D). However, only E-cadherin, but not other EMT-related genes, was significantly up-regulated by increased expression of miR-151-3p, suggesting that E-cadherin might be the major contributor in the TWIST1-promoting cell metastasis that is regulated by miR-151-3p. Moreover, increased expression of miR-151-3p did not affect cell growth and tumorsphere formation of breast cancer cells (Fig 4A, 4B and 4C). These findings suggest that microRNAs might target TWIST1 to regulate different cancer properties in different cancer types, and miR-151-3p might negatively regulate cell metastasis during breast cancer progression.
MDA-MB-231 cells (4 × 10 5 ) were transfected with empty vector (pEV) or the overexpression construct pFLAG-TWIST1 that contained the TWIST1 protein coding region (that lacks the TWIST1 3'UTR) (pTWIST1) for 48 hours. The cells (1 × 10 4 ) were seeded for transwell migration. After incubation for 12 hours, the migrated cells were counted from 10 random fields under a microscope. All experiments were performed in triplicate. Each bar represents the mean ± SEM. *, P < 0.05. doi:10.1371/journal.pone.0168171.g005 miR-151-3p Targets TWIST1 to Repress Breast Cancer Cell Migration The miR-151 gene resides within intron 22 of the FAK gene [16]. The data obtained from TCGA revealed that the miR-151 expression levels in breast cancers are positively correlated with mRNA levels of the miR-151 host gene FAK (Fig 2B). These results suggest that the increased miR-151 expression might be co-regulated with the FAK gene expression in breast cancers. It has been shown that p53 can bind to the FAK promoter and subsequently repress FAK expression. Recently, it was demonstrated that activation of the Notch1 pathway enhances FAK and miR-151 expression [23]. The Notch1 pathway may interact with p53 to regulate FAK and miR-151 expression levels. The mechanism underlying up-regulated expression of miR-151 in breast cancers requires further investigation.
FAK is a key signaling molecule that is activated by numerous stimuli, is often overexpressed in human tumors, and promotes cancer cell invasion and metastasis [35,36]. FAK and miR-151 have been found to promote cell migration and spreading of HCC [16,17] and gastric cancers [23]. Although miR-151 expression levels are positively correlated with the FAK mRNA levels in breast cancers (Fig 2B), our results revealed that ectopic expression of miR-151 represses cancer cell migration and invasion (Figs 4D, 4E and 5B), whereas knockdown of miR-151-3p enhances breast cancer cell migration (Fig 6D). These results indicate that unlike in HCC and gastric cancer, miR-151-3p may play a negative regulator in breast cancer metastasis.
Some direct targets of miR-151 have been identified. Downregulation of RhoGDIA by miR-151 was demonstrated to enhance HCC cell migration through activation of Rac1, Cdc42 and Rho GTPases [16]. It was also found that miR-151-3p represses the full-length isoform of human neurotrophin-3 receptor gene NTRK3 in neuroblastoma cells [37]. It was recently demonstrated that miR-151-5p and a combination of other miRNAs (miR-145a-5p or miR-337-3p) are able to significantly repress TWIST1translation and result in decreased migratory potential of murine embryonic fibroblast cells [19]. In the role of miR-151-3p in muscle fiber type determination, miR-151-3p was shown to directly target ATP2a2, a gene that encodes for a slow skeletal and cardiac muscle specific Ca 2+ ATPase known as SERCA2, thus downregulating slow muscle gene expression [38]. In the present study, we identified TWIST1 as the direct target of miR-151-3p. Our results clearly showed that ectopic expression of miR-151-3p represses TWIST1 expression (Figs 1C, 1D, 3A, 3B and 5A), whereas knockdown of miR-151-3p increases TWIST1 expression (Fig 6C). These results demonstrated that TWIST1 expression is negatively regulated by miR-151-3p. Interestingly, the data obtained from TCGA revealed a significant negative correlation between the TWIST1 mRNA levels and miR-151 expression levels in human breast cancers (Fig 3C). Moreover, miR-151 levels is significantly positive correlated with E-cadherin gene expression levels, and is significantly negative correlated with the gene expressions of fibronectin, osteonectin, and α5-integrin (Table 1). It has been reported that the expression E-cadherin is repressed by TWIST1, while the expressions of fibronectin, osteonectin, and α5-integrin are activated by TWIST1 [25]. These results suggest that miR-151-3p might regulate TWIST1 expression and thereby TWIST1-related downstream gene expression.
The data obtained from TCGA revealed that the miR-151 expression levels in breast cancer cells were higher than those in normal tissues (Fig 2A), and no differences were observed in were measured by miRNA quantitative real-time PCR. (C) The HCC1937 cells were transfected with 50 nM antogomir-151-3p (anti-miR-151-3p) or 50 nM scrambled control for 48 hours. TWIST1 and E-cadherin protein expression were measured by Western blot analysis. The data were normalized to α-tubulin protein levels. (D) HCC1937 cells were transfected with 50 nM antogomir-151-3p (anti-miR-151-3p) or 50 nM scrambled control for 48 hours. The cells (1 × 10 4 ) were seeded for transwell migration. After incubation for 12 hours, the migrated cells were counted from 10 random fields under microscope. All experiments were performed in triplicate. Each bar represents the mean ± SEM. *, P < 0.05.  (Fig 2C). Our present in vitro data suggest that miR-151-3p may negatively regulate the migration and invasion of breast cancer cells. The results are consistent with the findings that miR-151 expression was lower in the lymph-node metastases than in their corresponding tumors of breast cancer patients [18]. Recently, upregulation of miR-151 has been reported in esophageal cancer [39] and papillomavirus (HPV)associated oropharyngeal carcinoma [40]. The up-regulation of miR-151 is found to be associated with poor survival in human HPV-associated oropharyngeal carcinoma [40]. In cholangiocarcinoma, the concomitant dysregulation of both miR-151-3p and miR-126 were reported as factors related to the greatest improvement in overall survival [41]. In addition, an increase in miR-151-3p was identified as one of the most significant differences in miRNA expression for recurrent tumors compared to non-recurrent tumors of the patients with early stage lung adenocarcinomas [42]. Recently, miR-151-3p expression was found to be associated with the ductal carcinoma in situ (DCIS) subtype of breast cancers [43]. These findings suggest that the role of miR-151 in cancer progression seems to be tissue-or cancer type-specific. It is possible that the increased miR-151-3p in breast cancers could be partly involved in the regulation of TWIST1 gene expression at different stages of cancer progression. The TWIST1 expression in breast cancers may be regulated by several mechanisms, including the transcription factors in response to EMT stimulators and miRNAs. These findings need to be further verified using a xenograft model and using the clinical data for immunohistochemical staining of TWIST1 which show anti-correction between TWIST1 and miR-151-3p in breast metastasis.
In conclusion, our results suggested that TWIST1 is the direct target of miR-151-3p and that miR-151-3p may play a negative regulatory role in breast cancer metastasis. Further studies on the regulation of miR-151-3p expression in breast cancer may benefit the development of more specifically targeted therapies for patients with breast cancer metastasis.