miR-655 Is an EMT-Suppressive MicroRNA Targeting ZEB1 and TGFBR2

Recently, the epithelial-to-mesenchymal transition (EMT) has been demonstrated to contribute to normal and disease processes including cancer progression. To explore EMT-suppressive microRNAs (miRNAs), we established a cell-based reporter system using a stable clone derived from a pancreatic cancer cell line, Panc1, transfected with a reporter construct containing a promoter sequence of CDH1/E-cadherin in the 5′ upstream region of the ZsGreen1 reporter gene. Then, we performed function-based screening with 470 synthetic double-stranded RNAs (dsRNAs) mimicking human mature miRNAs using the system and identified miR-655 as a novel EMT-suppressive miRNA. Overexpression of miR-655 not only induced the upregulation of E-cadherin and downregulation of typical EMT-inducers but also suppressed migration and invasion of mesenchymal-like cancer cells accompanied by a morphological shift toward the epithelial phenotype. In addition, we found a significant correlation between miR-655 expression and a better prognosis in esophageal squamous cell carcinoma (ESCC). Moreover, ZEB1 and TGFBR2, which are essential components of the TGF-b signaling pathway, were identified as direct targets of miR-655, suggesting that the activation of the TGF-b-ZEB1-E-cadherin axis by aberrant downregulation of miR-655 may accelerate cancer progression.


Introduction
The epithelial-to-mesenchymal transition (EMT) is an essential biological process with remarkable morphological changes between the epithelial and mesenchymal states [1], and plays key roles in embryonic development, cancer and other diseases [2][3][4][5]. During the acquisition of EMT characteristics, cancer cells lose the expression of genes that promote cell-cell contact, such as Ecadherin and the miR-200 family, and gain the expression of mesenchymal markers, such as vimentin, fibronectin, and N-cadherin, leading to enhanced cancer cell migration and invasion [6][7][8] and to confer drug resistance characteristics on cancer cells [9]. Therefore, the development of EMT inhibitors may provide novel strategies for the prevention, diagnosis and treatment of cancers.
MicroRNAs (miRNAs) are endogenous small non-proteincoding RNAs of 19-22 nucleotides. These single-stranded RNAs are considered to play crucial roles in many normal cellular processes [10,11,12], and the multistep processes of carcinogenesis and cancer progression [13][14][15], depending on their specific gene targets. Furthermore, The many achievements in the field of the discovery of tumor-suppressive miRNAs (TS-miRNAs) and in vitro/in vivo delivery technology may offer the possibility of new therapeutic approaches for cancer. Since one miRNA can target an unpredictable number of messenger RNAs (mRNAs) of protein-coding genes on a genome-wide scale, the clinical applications of miRNAs for cancer therapies are considered better than those of short interfering RNAs (siRNAs). In addition, among miRNA-based approaches by in vivo delivery including the use of DNA plasmids or viral vectors, miRNA replacement therapy using double-stranded RNAs (dsRNAs) mimicking TS-miRNAs may be one of the most promising, offering hope for new cancer therapies [15,16].
Recently, the miR-200 family (miR-141, -200a, -200b, -200c, and -429) and miR-205 have been demonstrated as EMT-suppressive miRNAs directly targeting ZEB1 and ZEB2 [17]. The miR-200-ZEB1-E-cadherin axis has been clarified to be a crucial pathway downstream of TGF-b in EMT while reciprocal repression between ZEB1 and the miR-200 family has recently been reported to promote EMT and invasion in cancer cells [18][19][20][21][22]. Actually, EMT-induced cancer cells were also reported to be more efficient at forming cancer stem cells with invasive and tumorigenic phenotypes [23]. Therefore, EMT-suppressive miRNAs in cancers have been considered to be important diagnostic markers and new therapeutic agents for human malignancies.
Herein, we show the identification of a novel EMT-suppressive miRNA by function-based screening using 470 synthetic miRNAs and the detailed characterization of the miRNA and its direct targets. The function-based screening makes it possible to analyze the biological effects of a large number of dsRNAs on cancer cells directly. In addition, this approach has already proved successful in the exploration of dsRNAs having oncogenic or tumorsuppressive effects on cancer cells [24][25][26][27]. In the present study, to detect the promoter activity of CDH1/E-cadherin by measuring the fluorescence intensity of ZsGreen1 protein in our functionbased screening, we established a unique cell-based reporter system using a pancreatic cancer cell line, Panc1, having phenotypic plasticity at EMT/mesenchymal-to-epithelial transition (MET). The present study is the first to show clearly that miR-655 targets ZEB1 and TGFBR2 inducing inactivation of the TGFb signaling pathway, involving the miR-200-ZEB1-E-cadherin axis, strongly suggesting a potential role for miR-655 as a prognostic marker and therapeutic agent in human cancers.

Cell Lines and Primary Tumor Samples
The culture conditions for the pancreatic cancer [28], esophageal squamous-cell carcinoma (ESCC) [29] and oral squamous cell carcinoma (OSCC) [30,31] cell lines were reported previously. These cell lines were authenticated in previous studies with array-based comparative genomic hybridization (aCGH) analyses [28,29]. A breast cancer cell line, MDA-MB-231, was purchased from the American Type Culture Collection (Manassas, VA, USA) and maintained in the medium recommended by the manufacturers. Primary ESCCs and OSCCs were obtained with the written consent of each patient after approval by a local ethics committee of Medical Research Institute and Faculty of Medicine, Tokyo Medical and Dental University (Approval ID: 2010-5-2).
Transfection with Synthetic miRNAs and Small Interfering RNAs (siRNAs) 10 nM of dsRNA mimicking human mature miRNA or control non-specific miRNA (Ambion, Austin, TX; Thermo Scientific Dharmacon, Lafayette, CO) was transfected individually into cells using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA). The function-based screening was performed using Pre-miR TM miRNA Precursor Library-Human V3 (Ambion) in duplicate [26,27]. The numbers of viable cells were assessed by the colorimetric water-soluble tetrazolium salt (WST-8) assay (Cell counting kit-8; Dojindo Laboratories, Kumamoto, Japan). The reporter construct was generated using the pZsGreen1-1 Vector (Clontech Laboratories, Palo Alto, CA). The fluorescence intensity of the ZsGreen1 protein was measured by ARVO mx (Perkin Elmer, Waltham, MA).

Transwell Migration and Invasion Assay
Transwell migration and invasion assays were carried out in 24well modified chambers precoated with (invasion) or without (migration) Matrigel (BD BioCoat, BD Biosciences, Franklin Lakes, NJ) as described elsewhere [32]. Cells in serum-free medium were transferred into the upper chambers. After incubation, the cells that migrated into the lower chambers with 10% FBS as the chemoattractant were fixed and stained with the Diff-Quik stain (Sysmex, Kobe, Japan), and counted in 5 random fields. Each assay was performed in triplicate.

Real-time Reverse Transcription-PCR and miRNA Target Predictions
Real-time reverse transcription-PCR (RT-PCR) was performed as described elsewhere [33]. Predicted targets for miRNAs and their target sites were analyzed using microRNA.org. All samples were analyzed in a duplicated manner.

Statistical Analysis
The association between clinicopathological characteristics and status of miR-655 expression in ESCC patients was evaluated with x 2 or Fisher's exact test (Table S2). A p-value less than 0.05 was defined as being statistically significant. In Kaplan-Meier curves, differences between subgroups were tested with the log-rank test. Differences between subgroups were tested with the Mann-Whitney U test.

Establishment of a Cell-based Reporter System for Investigating CDH1/E-cadherin-promoter Activity in Panc1 Cells
To perform function-based screening of EMT-suppressive miRNAs, we established a cell-based reporter system. The promoter region of the CDH1/E-cadherin gene located in the 59 untranslated region (59UTR) and Exon1 (from nt 21001 to +57 relative to the transcription start site, TSS, at the 59end of the gene) was prepared by genomic PCR using specific primers (Table  S1), and was inserted into the pZsGreen1-1 vector at the multiple cloning site upstream of the promoterless ZsGreen1 gene (Fig. 1A) [34]. We first transfected the above-mentioned construct into Panc1 cells, the phenotypic plasticity of which had already been confirmed in previous studies [35,36], and then carried out cloning of stable transfectants by limiting dilution. Furthermore, among many clones, the most-reactive single-cell clone, PEcadZsG-Panc1, was selected by measuring the fluorescence intensity of the ZsGreen1 protein induced by transient transfection of miR-200a or -200b, already known as EMT-regulating miRNAs. In this clone 96 hours after transfection with 10 nM of dsRNA mimicking miR-200a or -200b, the fluorescence intensity of ZsGreen1 was remarkably increased as compared with that in the control counterparts (Fig. 1B). RT-PCR and Western blot analyses also showed a consistent correlation between the fluorescence intensity of ZsGreen1 and expression levels of CDH1/E-cadherin mRNA and protein in these transfectants. Consequently, we judged that ZsGreen1 expression was tightly regulated under the CDH1/Ecadherin-promoter in our cell-based reporter system for functionbased screening of EMT-suppressive miRNAs.

Function-based Screening of EMT-suppressive miRNAs with our cell-based Reporter System
To identify EMT-suppressive miRNAs, we performed functionbased screening, in which the fluorescence intensity of ZsGreen1 was made an index, using our cell-based reporter system and 470 dsRNAs at 10 nM. Figure 1C and Table S2 demonstrate results of this screening in a Panc1 stable clone, PEcadZsG-Panc1, 96 hours after transient transfection with each dsRNA. In Table 1, 17 miRNAs, the relative fluorescence intensity of which remarkably increased in our screening (.2.2-fold change of mean fluorescence intensity compared with the control counterpart), were enrolled as candidate EMT-suppressive miRNAs. Among these candidates, we excluded well-known EMT-suppressive miRNAs, such as miR-200a and -200c [8,21,37,38], and selected 10 miRNAs (miR-96-5p, -132-3p, -183-5p, -139-5p, -217, -520d-3p, -526b-3p, -629-3p, -655, and -200b-3p) showing a consistent positive correlation between two sets of fluorescence data taken from the fluorescence microplate reader (Table S2) and fluorescence microscope (Fig.  S1). In a Western blot analysis in the parental Panc1 cells 96 hours after transient transfection with these miRNAs, miR-520d-3p and miR-655, as well as miR-200b-3p, were confirmed to upregulate expression of the E-cadherin protein markedly, whereas only slight effects of other miRNAs were observed (Fig. 1D). Moreover, the relative fluorescence intensity of miR-655 was clearly higher than that of miR-520d-3p (Table 1), suggesting miR-655 to be a prime candidate for EMT-suppressive miRNA. and Western blot analysis (Right) for expression of the CDH1/E-cadherin transcript and protein, respectively, in these transfectants. C, Results of the function-based screening of EMT-suppressive miRNAs in a cell-based reporter system using Pre-miR TM miRNA Precursor Library-Human V3 (Ambion) containing 470 dsRNAs mimicking human mature miRNAs. The fluorescence intensity of ZsGreen1 was evaluated by fluorescence microplate reader in duplicate. The relative fluorescence intensity in each transfectant was calculated by normalization of each result to the fluorescence intensity in control cells transfected with non-specific miRNA (see Table 1 and Table S2). The lower closed arrow indicates the 470 miRNAs examined. D, Western blot analysis of E-cadherin protein levels in parental Panc1 cells 96 hours after transient transfection with 10 nM of ds-NC or 10 nM of ds-miRNAs mimicking miR-96-5p, -132-3p, -183-5p, -139-5p, -217, -520d-3p, -526b-3p, -629-3p, -655 and -200b-3p. Because miR-200b has already been confirmed to induce expression of the CDH1/E-cadherin transcript and protein in this study (Fig. 1B) Fig. 2A and Fig. S2). We noticed a consistent positive correlation among expression profiles of miR-200 family members and a slight correlation between CDH1/E-cadherin and miR-200 family members. Although no correlation between the expression pattern of miR-655 and that of any of these marker genes was found in this panel, these cell lines all showed lower expression of miR-655 than the miR-200 family and miR-205, as compared with a normal pancreas. Marked down regulation of miR-655 expression was also observed in 97.7% (42/43) of ESCC and 94.7% (18/19) of OSCC cell lines (,0.5-fold expression) (Fig. S3A  and B). Moreover, the expression of endogenous miR-655 was higher in MCF7 and MCF10A (human breast epithelial cells) than MDA-MB-231, suggesting that downregulation of miR-655 might contribute to phenotypic stabilization of mesenchymal feature in MDA-MB-231 cell line (Fig. S3C).
To confirm the EMT-suppressive effects of miR-655 on mesenchymal-like pancreatic or breast cancer cells having phenotypic plasticity at EMT/MET, we ectopically introduced 10 nM of synthetic dsRNA mimicking mature miR-655 into Panc1, KP1N, KP4-4 and MDA-MB-231 cells. The Panc1 and KP1N cell lines are miR-655-high expressers, while the KP4-4 and MDA-MB-231 cell lines are miR-655-low expressers ( Fig. 2A). However, in all four cell lines 96 hours after transfection with miR-655, a morphological shift toward the epithelial phenotype was induced (Fig. 2B) consistent with an upregulation of CDH1/E-cadherin expression at the mRNA and protein levels (Fig. 2C). In addition, we confirmed that ectopic expression of miR-655 increased CDH1/E-cadherin expression at the mRNA and protein levels in an ESCC cell line, TE8, and an OSCC cell line, HSC2, (Fig. S4), although a morphological shift toward the epithelial phenotype in these cell lines was not observed (data not shown). These EMT-suppressive effects of miR-655 were observed in a miR-200 family-independent manner (data not shown). To take into consideration off-target effects of dsRNAs, these EMT-suppressive effects of miR-655 were also confirmed using two kinds of dsRNAs purchased from independent companies (Fig. S5). Notably, effects of overexpression of exogenous miR-655 on cell growth were not constant in these cell lines (Fig. 2D), whereas the number of cells that migrated through the uncoated or Matrigel-coated membranes in cell migration or invasion assays, respectively, was significantly decreased in all miR-655-transfectants compared with their control counterparts ( Fig. 2E and 2F). These results suggest that miR-655 may suppress EMT in mesenchymal-like cancer cells.

Expression Analysis of miR-655 in Primary ESCC and OSCC Cases
We investigated the normal human tissue distribution and tumor expression of endogenous miR-655 by TaqMan RT-PCR analysis. Among 22 normal tissues, upregulation of miR-655 expression was observed in brain, cervix, esophagus and placenta (.2-fold increase compared with a normal pancreas) (Fig. 3A). We next examined the expression level of the miR-655 transcript in primary tumors of ESCC and OSCC, respectively. Expression levels of miR-655 in tumors as compared with paired non-tumorous mucosae were markedly reduced in 44.8% (13/ 29) and 60.9% (14/23) of primary ESCC and OSCC cases, respectively (,0.5-fold expression) (Fig. 3B). Furthermore, to evaluate the clinical significance of miR-655 expression in ESCC, we categorized the patients into two groups based on the mean value: a low miR-655 group (n = 18) and a high miR-655 group (n = 11). In Kaplan-Meier survival curves for 29 patients with ESCC expressing higher and lower levels of miR-655, univariate analyses of overall and non-recurrent survival with log-rank tests demonstrated a significant association between higher levels of miR-655 expression and a better survival rate (Fig. 3C, P = 0.0359, log-rank test), whereas the miR-655 expression in each tumor was not associated with clinicopathological features (Table S3). These findings suggest that miR-655 expression may significantly correlate with prognosis in ESCC. We could not analyze the prognostic significance of miR-655 expression in OSCC because complete survival data was not included in our clinical data. Finally, we analyzed miRNA-target associations at the mRNA level in ESCC and OSCC primary samples (Fig. S6A), however significant correlations were not found between miR-655 expression and expression of ZEB1 or TGFBR2 transcripts as well as a large number of miRNAs and their targets, indicating that not only miR-655 but also other unknown molecules including transcription factors might regulate ZEB1 and TGFBR2 expression. Similar to result in primary samples, there were no correlations between miR-655 expression and mRNA or protein expression of these targets in ESCC cell lines because of remarkable reduction of miR-655 expression in almost ESCC cell lines (Fig. S6B).

Characterization of ZEB1 and TGFBR2 as Novel Direct Targets of miR-655
We searched the websites microRNA.org (http://www. microrna.org/) and Target Scan Human 6.2 (http://www. targetscan.org/) [39,40] for direct targets of miR-655, and focused on ZEB1 and TGFBR2 as potential candidates, respectively. Although expression levels of ZEB1 and TGFBR2 tended to be lower in pancreatic cancer cell lines and a breast cancer cell line relative to a normal pancreas (Fig. 4A), transcript and protein levels of these candidate genes were markedly reduced in mesenchymal-like pancreatic or breast cancer cells 96 hours after transfection with dsRNAs mimicking miR-655 (Fig. 4B). We obtained the same results using other sets of dsRNAs purchased from independent sources (Fig. S7). Moreover, in the luciferase reporter assay with vectors containing the wild type or a mutated 39-UTR target site of ZEB1 (region 4) and TGFBR2 (region 1) downstream of the luciferase reporter gene, we detected statistically significant reductions in luciferase activity in wild type constructs, but not in mutant constructs (Fig. 4C and Fig. S8A and B), indicating that ZEB1 and TGFBR2 were novel direct targets of miR-655. We confirmed that the treatment with TGF-b could induce EMT accompanied by the upregulation of TGFBR2, TGFBR1, SNAI1/Snail, ZEB1, and PAI-1 and the downregulation of CDH1/E-cadherin at mRNA and protein levels in KP1N cells. Furthermore, overexpression of exogenous miR-655 significantly inhibited the upregulation and downregulation of these EMTregulatory genes in KP1N cells treated with and without TGF-b, respectively ( Fig. 4D and Fig. S9), whereas miR-655 could not change levels of phosphorylated Smad2/3 in these cells. Our findings strongly suggest that the TGF-b-induced EMT can be suppressed by miR-655, independently of miR-200 family members, through translational inhibition of ZEB1 and TGFBR2 in cancer cells (Fig. 4E).

Discussion
EMT plays a crucial role in many stages not only in embryonic development but also in cancer progression [3,6]. Cancer cells undergoing EMT are endowed with more aggressive phenotypes, such as mesenchymal and stem cell-like features, resulting in the acquisition of malignant properties, such as invasion, metastasis, recurrence, and drug resistance [4,5,41]. The evidence for EMT, including our own [32,35], led us to consider that the development of EMT inhibitors might provide opportunities for both prevention and treatment of cancer. Therefore, to identify EMTsuppressive miRNAs, we performed here the function-based Over the last few years, we have focused on miRNAs as key post-transcriptional regulators of gene expression and previously identified four tumor-suppressive miRNAs (TS-miRNAs) directly targeting oncogenic genes in oral squamous cell carcinoma (OSCC) and hepatocellular carcinoma (HCC) using expression-based and DNA methylation-based screening, respectively [15,33,42]. In recent studies, we successfully performed functionbased screening with a cell proliferation assay for 327 synthetic miRNAs and identified two TS-miRNAs directly targeting Rictor in OSCC and endometrial cancer (EC) [15,26,27]. In the present study, a unique cell-based reporter system for investigating CDH1/ E-cadherin-promoter activity in the Panc1 cell line was established Figure 3. Expression analysis of miR-655 in primary ESCC and OSCC cases. A, TaqMan real-time RT-PCR analysis of endogenous miR-655 in 22 normal human tissues (Ambion and Clontech). Marked upregulation of miR-655 expression (.2-fold increase compared with pancreas) was observed in brain, cervix, esophagus and placenta. B, Expression profiles of miR-655 in a panel of paired tumorous and non-tumorous tissues from primary ESCC and OSCC cases. Bar graphs show the ratio of the expression level in tumors (T) to those in their paired normal mucosae (N). C, Kaplan-Meier survival curves for high and low miR-655 groups based on TaqMan real-time RT-PCR. In univariate analyses of overall and non-recurrent survival with log-rank tests, a high level of miR-655 expression was significantly associated with a much better survival rate among patients with ESCC (P = 0.0359, log-rank test). doi:10.1371/journal.pone.0062757.g003 and used in the function-based screening of EMT-suppressive miRNAs. In this system, the CDH1/E-cadherin-promoter was employed to monitor MET, because the transcription of CDH1/ E-cadherin is known to be repressed during EMT and activated during MET [26]. Panc1 has already been used as an in vitro experimental model for assessing induction of EMT [35,36]. Since Panc1 was also confirmed to be a mesenchymal-like cell line with phenotypic plasticity at EMT/MET [35], we used it in the establishment of our cell-based reporter system. Actually, miR-200a, -200b and -200c were discovered as EMT-suppressive miRNAs in previous studies, emerged at the top of a list of results of screening, suggesting the present cell-based reporter assay to be a powerful tool for high-throughput function-based screening of miRNAs, siRNAs and chemical compounds having EMTsuppressive effects.
Here, we successfully identified miR-655 for the first time as a novel EMT-suppressive miRNA through function-based screening. This miRNA had mostly remained uncharacterized in the field of cancer research. The miR-655 gene is located within a noncoding region at 14q32.31, which harbors 50 intergenic miRNA genes within a limited region of 198 kb. Although notable copy number aberrations were not detected at 14q32.31 by our aCGH analyses using a panel of pancreatic cancer [28], ESCC [29] and OSCC [30,31] cell lines, the expression of miR-134 and miR-370 located at this locus was described to be significantly lower in gastrointestinal stromal tumors (GISTs) with 14q loss and also in GISTs with tumour progression [43]. In the present study, the expression of miR-655 was largely downregulated in a panel of pancreatic cancer, ESCC and OSCC cell lines, and a breast cancer cell line, MDA-MB-231. Moreover, we found a significant correlation between higher levels of miR-655 expression and a better survival rate in patients with ESCC, suggesting miR-655 expression to be a promising prognostic marker for ESCC.
In the present study, ZEB1 and TGFBR2 were identified as direct targets of miR-655. These targets have been revealed to be major components of TGF-b signaling pathways and to induce EMT through repression of CDH1/E-cadherin [17,19,44]. ZEB1 was identified first as a strong predictor of poor survival and distant metastasis in colorectal adenocarcinoma, breast cancer [45,46], and lung adenocarcinoma [47]. High TGFBR2 expression was also correlated with a shorter overall survival in estrogen receptor-negative breast cancer [48]. miR-520c, miR-373, and miR-211 were described as miRNAs targeting TGFBR2 and contributing to the induction of MET [25,49] although these three miRNAs were not identified as prime candidates of EMTsuppressive miRNA in our screening. The miR-200 family members are typical EMT-suppressive miRNAs targeting several components of TGF-b signaling pathways, including the miR-200-ZEB1-E-cadherin axis, which is crucial in EMT and was described to be deregulated in mesenchymal-like cancer cells [17,18,20]. Although the miR-200 family and miR-205, like miR-655, target ZEB1, their biological functions were found to differ from those of miR-655. First, besides ZEB1, TGFBR2 was characterized as a direct target of miR-655 in our study, but not the miR-200 family and miR-205. Second, our studies past and present, have showed the mesenchymal-specific downregulation of miR-200 expression in a panel of OSCC [35] and pancreatic cancer cell lines, respectively, but not miR-205 and miR-655 expression. These differences between miR-655 and miR-200 family members indicate the biological function of each EMT-suppressive miRNA in physiological and pathophysiological processes, including EMT/MET. In addition, several components of TGF-b signaling pathways, TGFBR2, Snail and ZEB1, were reduced directly or indirectly by overexpression of miR-655 in cancer cells treated with or without TGF-b. Recent studies have demonstrated that a miRNA significantly decreased signal output over time, by reducing the concentration of several components in a signaling cascade [50,51]. These results strongly support our findings that a single EMT-suppressive miRNA may target several EMTinducible components of a specific signaling pathway and coordinate their expression. On the other hand, overexpression of miR-655, as well as the miR-200 family, induced significant morphologic changes and inhibited cell migration and invasion in 3 pancreatic cancer cell lines and a breast cancer cell line, MDA-MB-231. These observations suggest the EMT-suppressive effects of miR-655 to be essential for cancer progression.
In conclusion, we established a unique cell-based reporter system for monitoring the promoter activity of CDH1/E-cadherin. By using the system for the first time we identified miR-655 as a novel EMT-suppressive miRNA, the biological meaning of which was different from that of the miR-200 family. Overexpression of miR-655 remarkably increased E-cadherin expression and suppressed cell motility in several cancer cell lines, clearly indicating that this miRNA is a strong suppressor of EMT. In ESCC, miR-655 expression demonstrated a significant association with a better prognosis. Furthermore, ZEB1 and TGFBR2, which are cardinal components of the TGF-b signaling pathway, were characterized as direct targets of miR-655. Our results suggest the potential of the EMT-suppressor miR-655 targeting ZEB1 and TGFBR2 as a prognostic marker and therapeutic agent for cancer. Figure S1 Fluorescence micrographs of a stable Panc1 clone 96 hours after transient transfection with dsRNA in functional-based screening using Pre-miR TM miRNA Precursor Library -Human V3 (Ambion). The Panc1 clone was established by transfection with a reporter construct containing a promoter sequence of CDH1/E-cadherin in the 59 upstream region of the ZsGreen1 reporter gene and cloning using limiting dilution (see Fig. 1A, and 1B). Each dsRNA was transfected individually into the clone. These 17 miRNA genes were selected as candidates for EMTsuppressive miRNAs in functional-based screening (see Table 1, Fig. 1C, and Table S2)  Table S2 Summary of functional-based screening using 470 dsRNAs mimicking mature miRNAs (Pre-miR TM miRNA Precursor Library -Human V3, Ambion) and a stable Panc1 clone transfected with a reporter construct containing a promoter sequence of CDH1/E-cadherin in the 59 upstream region of the ZsGreen1 reporter gene. (XLS)