Absence of miR-378d Promoted The Malignant Phenotype of ESCC Through The AKT-β-Catenin and Hippo-p53 Signaling Pathway

Background: Chemoresistance is an important cause of malignant progression of esophageal squamous cell carcinomas (ESCCs). miR-378d is sharply reduced in paclitaxel (PTX)-resistance esophageal cancer cells by gene-expression prole analysis (RNA-Seq), but the mechanism of miR-378d-mediated tumor progression is unclear. Patients and methods: Herein, we detected miR-378d expression in 596 ESCC patients by in situ hybridization. Results showed that low miR-378d expression was associated with poor prognosis of ESCC patients, and that miR-378d absence enhanced carcinogenesis by promoting chemoresistance, colony formation, EMT, invasion, and metastasis. Results: Furthermore, miR-378d can target downregulated AKT1 expression by binding to the AKT1 mRNA 3′UTR, inactivating the AKT-β-catenin signaling pathway, and reducing the epithelial–mesenchymal transition marker Vimentin and the cancer stem cell marker ALDH1A1. miR-378d silencing in ESCC cells also promoted polyploidy formation in vitro and in vivo, and miR-378d inhibition suppressed the Hippo-p53 signaling pathway. Consequnetly, YAP and TAZ protein accumulated in nuclei and p53 expression decreased, which may promote the formation of ploidy tumor cells. Conclusions: Therefore, low miR-378d expression is a poor prognostic factor of ESCC patients and promotes polyploidy and cancer progression by activating AKT-β-catenin and suppressing the Hippo-p53 signaling pathway. β-catenin

(TNM Classi cation of Malignant Tumours, Eighth edition). The study protocol had been reviewed and approved by the local ethics committee. All patients gave written consent for their tissue samples. This research was approved by the ethics committee of Jining Medical University. Each patient signed an informed consent form.

Tissue microarray
Representative areas of the ESCC were marked on each hematoxylin-eosin (H&E) slide and tissue para n block, and the marked areas of tissue para n blocks were sampled for TMAs. TMAs were assembled with a tissue-arraying instrument (Beecher Instruments, Silver Springs, MD, USA) as described by Kallioniemi et al. [22]. .

In Situ Hybridization
ESCC TMA was dewaxed in xylene, rehydrated in alcohol gradient, and washed two times with DEPC-PBS.
The sections were treated with 2 μg/mL proteinase K (Roche) for 15 min at 37 °C and washed three times with DEPC-PBS. Then, the sections were acetylated 15 min at room temperature (acetic anhydride in DEPC-water, 6 N HCl, and triethanolamine) and subsequently washed three times with DEPC-PBS.
Sections were prehybridized in hybridization buffer (50% formamide; 5× saline sodium citrate; pH 7.0; 100 μg/mL sheared salmon sperm DNA, 0.5 mg/mL yeast tRNA, and 1× Denhardt's solution) at 58 °C for 1 h before the buffer was replaced with hybridization solution containing miR probe. The miR-378d detection probes labeled with digoxin at 5′-end was from Boster (#MK10502). Probes were diluted in prehybridization buffer to a concentration of 5 nM and hybridized with the sections overnight at 58 °C according to the RNA melting temperature of probes. After hybridization, the sections were washed three times with 2 × SSC and 0.2 × SSC, permeabilized for immunostaining with 0.1% Triton X-100, and washed two times with PBS. Unspeci c background was blocked with 5% swine serum diluted in PBS/BSA for 30 min.

PTX treatment
All cell lines were cultured in complete medium until the cells reached 90% con uence. Different concentrations of PTX were added to the different cells, which were then treated for 24 h. PTX was then withdrawn, the medium was replaced, and the cells were cultured until no signi cant cell death was observed.

Sequencing of miRNA and microarray analysis
The small RNA of TE-1 control and TE-1-PTX (9 days) treated with PTX was used for miRNA sequencing.
The miRNA-sequencing libraries were constructed according to the protocol for the Illumina small RNA sample preparation kit. Sequencing was performed on an Illumina HiSeq 2000 sequencer. Library construction and sequencing were performed at Genergy Biotech (Shanghai). miRNA expression was analyzed with miRdeep2.0.0.7 [23]. , and differentially expressed miRNAs were identi ed using an FDR cutoff value of 0.05. mRNA expression pro ling was conducted with Roche NimbleGen Human 12 × 135 K Gene Expression Array by KangChen Bio-tech. Raw data were processed with RMA algorithm, and differential expression analysis was performed with R package limma37 (Version 3.22.7).

Cell transfection
Transfection of plasmids was performed using Lipofectamine TM 3000 reagent (Invitrogen, USA) according to the manufacturer's instructions. Transfection of miRNA mimics or inhibitors (Ribobio, China) was performed using Lipofectamine RNAiMAX (Invitrogen, USA) at a nal concentration of 20 nM.

Lentivirus packaging and transduction
Vectors were packaged in 293FT cells using ViraPower Mix (Genepharma). After culturing for 48 h, lentiviral particles in the supernatant were harvested and ltered by centrifugation at 500 g for 10 min, and transfected into ESCC cells. The cells were then cultured under puromycin (10 μg/mL) selection for 2 weeks, after which real-time PCR was used to determine the level of miR-378d. Cell lines stably expressing miR-378d-inhibitor or negative control (NC) vector were designated as Lv-miR-378d-inhibitor and Lv-miR-NC cells, respectively.

Transwell migration and invasion assay
In vitro cell migration assay was performed using transwell chambers (8 mm pore size; Corning). Cells were plated in serum-free medium (2 × 10 4 cells per transwell). Medium containing 15% FBS in the lower chamber served as a chemoattractant. After 48 h, the nonmigrating cells were removed from the top face of the lters by using cotton swabs, and the migratory cells located on the bottom side of the chamber were stained with crystal violet, air dried, photographed, and counted. Images of ve random elds at 10× magni cation were captured from each membrane, and the number of migratory cells was counted.
Similar inserts coated with Matrigel were used to determine the cellular invasive potential in the invasion assay.
Matrigel 3D cell culture Cells (5×10 3 /50 μL) were seeded onto 96-well plates with a round-bottom lid made of ultralow attachment polystyrene (#7007, Costar, USA). The cells were cultured overnight and found to form one sphere per well. After discarding the medium and adding 75 μL of melted Matrigel (BD, USA) to resuspend the cell sphere, the mixture was incubated for 30 min for settling. Finally, 200 μL full medium/well was added, and the medium was changed every other day.

Colony-formation assays
Cells were seeded onto six-well plates (5 × 10 2 cells per plate) and cultured for 10 days. The colonies were stained with 1% crystal violet for 30 s after xation with 10% formaldehyde for 15 min and then imaged using the camera of an iPhone 5S (Apple, Inc., Cupertino, CA, USA).

Dual-luciferase reporter assay
In a typical procedure, 293T cells (3×10 4 cells per well) grown on a 24-well plate were co-transfected with luciferase reporter miRGLO-AKT1-3′UTR plasmid (WT or mutation type; Genepharma, Shanghai, China) (200 ng per well) and miR-378d (20 nM) using Lipofectamine™ 3000 (Invitrogen, USA). About 24 h later, a dualluciferase reporter assay kit (Promega, USA) was used to measure the luciferase and renilla activity of these samples according to the manufacturer's instructions.

F-actin cytoskeleton uorescence staining
Cells were grown on laminin-coated glass cover slips, xed in 4% paraformaldehyde, and stained with Phalloidin (Molecular Probes, USA). Cells were observed using a uorescence microscope (Leica, Germany).

Cell-Viability Assay
CCK8 was used to assess cell viability. KYSE510 and TE-1 cells (1 × 10 4 ) were seeded onto a 96-well plate in quintuplicate per well. About 12 h later, the cells were incubated with a gradient concentration of therapeutic drugs for 48 h. The medium was removed, RPMI1640 (90 μL) and CCK8 (10 μL) were subsequently added to each well, and the mixture was incubated for 3 h at 37 °C. A microplate reader was used to measure the optical density (OD) at 450 nm. The degree of drug response for tumor cells was estimated by dividing the half-maximal inhibitory concentration (IC50).

Liver transplantation
The animal protocol was approved by the ethical review committee of the A liated Hospital of Jining Medical University. Eight-week-old male BALB/c nude mice (Beijing Weitong Lihua Laboratory Animal Technology Co., Ltd.) were anesthetized with 4% chloraldehyde hydrate (100 μL/10 g), and body temperature was maintained by heating blankets. The thoracoabdominal skin of nude mice was sterilized with 75% alcohol and iodophor, respectively. After cutting open the skin at the lower right of the cartilago to expose the lobe of the liver, cell suspension (50 μL/25 μL serum-free and 25 μL of Matrigel containing 500 000 cells) was injected into the liver capsule slowly. Then, the syringe was pulled out and the injection port was pressed for 2-5 min with an iodophor cotton ball. Finally, the incision was sutured layer by layer according to the anatomical structure. Mice were sacri ced 44 days after tumor-cell inoculation. Afterwards, liver tissues, lung tissues, and abdominal-metastasis tumors were xed in 4% saline-buffered formalin, embedded in para n, sectioned at 4 μm, and stained with H&E and IHC.

Statistical Analysis
Statistical analyses were performed using the SPSS 13.0 software package (SPSS, Chicago, IL, USA) and GraphPad Prism Software (version 6, La Jolla, CA, USA). For statistical comparison of two groups, twosided Student's t test with the same variances was used. Differences between variables were analyzed by two-tailed or Fisher exact tests. Survival curves were plotted using the Kaplan-Meier method and compared with log-rank tests. Multivariate survival analysis was performed for all parameters found to be signi cant in univariate analysis using a Cox regression model. Comparisons between groups for statistical signi cance were performed with a two-tailed Student t test. Data are presented as the mean ± SD. P values < 0.05 were considered signi cant.
Results miR-378d showed reduced expression in ESCC cells after PTX treatment PTX is the rst-line drug for esophageal cancer, but drug resistance remains a problem. In this study, we used PTX to treat TE-1 cells. From days 2 to 9, the cells died continuously, and only a small number of cells remained at day 9. These cells slowly proliferated until day 20, the remaining cells began to proliferate rapidly, and the culture bottles were covered at day 24 (Fig. 1A). The remaining PTX-treated cells at day 9 (TE-1-PTX) and DMSO-treated cells (TE-1-NC) were used to detect miRNA differential expression by sequencing. miRNA gene-expression pro le data showed that miR-378d was signi cantly downregulated in remaining cells (Fig. 1B). In situ hybridization (ISH) assay con rmed that miR-378d was signi cantly downregulated in remaining PTX-treated cells, and that miR-378d was expressed in the cytoplasm and nucleus (Fig. 1C). Our data suggested that miR-378d absence may play an important role in PTX resistance.
Loss of miR-378d expression corresponded to poor prognosis We detected miR-378d expression by ISH in a ESCC tissue array containing 610 ESCC samples because the incompleteness of the para n section and antigen repair led to tissue shedding. Finally, 596 specimens were obtained. The data showed that miR-378d positive staining was primarily in the cytoplasm of squamous epithelial ( Fig (Table 1). However, our data showed that miR-378d expression was signi cantly negatively associated with the overall survival rate ( Fig. 2D; p < 0.0001) of ESCC patients.

Loss of miR-378d expression promoted the invasion and metastasis of ESCC
We then detected the stress ber formation and polymerized actin in miR-378d silencing cells and control cells using phalloidin staining. Results showed that miR-378d silencing cells underwent morphological changes, including the formation of an obvious actin lament cytoskeleton and an expansive morphological structure compared with control TE-1 and KYSE-510 cells (Fig. 4a). In vitro invasion assays using Matrigel-coated membranes revealed that the invasive potential of miR-378d cells was reduced compared with that of controls ( Fig. 4b; P < 0.001). Results of migration assay with Matrigelcoated free membranes were consistent with those of the invasion assays ( Fig. 4b; P < 0.001). These ndings suggested that miR-378d inhibited EMT. Colony-formation assay further showed that miR-378d silencing signi cantly enhanced the monoclonal colony-formation ability in TE-1 and KYSE-510 cells ( Fig.  4c; P = 0.0004 and P = 0.0003). Tumor-sphere-formation assay showed that 5000 ESCC cells aggregated into sphere culture overnight for ESCC cells, and that loss of miR-378d expression induced cell shedding from spheres at day 4 (Fig. 4d). We also found that miR-378d lost expression promoted the invasion of sphere in Matrigel (Fig. 4e).
Liver is a common metastasis site of esophageal cancer, so liver-transplantation assay was performed to analyze the metastatic ability of tumor cells without miR-378d expression. KYSE-150-miR-NC and KYSE-150-miR-378d-inhibitor cells were transplanted into subcapsular liver of BALB/c nude mice (hereafter denoted as NC-mice and In-mice, respectively. The experiment was terminated after 44 days. The body weight of In-mice was signi cantly lower than that of NC-mice at day 44 (P = 0.011, Fig.s 4f & g). Before the end of the experiment, NC-mice did not die (0/6), but two IN-mice (2/6) died at days 35 and 38, respectively (Fig. 4h). Although the number of liver nodules in In-mice was more than that in NC-mice, no statistical difference (Fig.s 4i & k) was observed. Tumor cells basically metastasized to the abdominal cavity, and no metastasis was found in all lung tissues. Abdominal metastatic tumors appeared in 5/6 of the NC-mice and 6/6 of the In-mice (Fig. 4j). The number of abdominal-metastasis tumors in In-mice was more than that in NC-mice ( Fig. 4l, P = 0.026), and the volume was also larger than that of the NC-mice. These data showed that loss of miR-378d expression promoted tumor metastasis.
The protein expression levels of β-catenin and the downstream target genes vimentin and ALDH1A1 decreased in cells with highly expressed miR-378d (Fig. 5h), whereas absent miR-378d expression increased these levels (Fig. 4i). The data showed that miR-378d absence activated the β-catenin pathway.
Loss of miR-378d expression promoted polyploidy We found that remaining PTX-treated cells were mostly large cells (Fig. 1a, day 9; Fig.s 1c & 6a, day 9). The polyploid cells (≥ 5N) signi cantly increased after PTX treatment (Fig. 6a). Circulating tumor cells (CTCs) are closely related to tumor metastasis [24]. and were usually de ned as tumor ploidy ≥ 3N. In the blood of ESCCs, we detected the CTCs of different ploidy cells (2N, 3N, 4N, and ≥ 5N) (Fig. 6b). We also analyzed the ploidy cells' proportion in ESCC and found that 3N was the most abundant (68.5%), followed by 4N (25.6%) and ≥ 5N (5.9%) (Fig. 6c), 2N tumor cells were rare and hard to distinguish from leucocytes in ESCC CTCs. DNA ploidy analysis showed that miR-378d silencing promoted polyploidtumor-cell formation in TE-1and KYSE-150 cells (Fig. 6d). Then, we detected polyploid-tumor-cell formation after PTX treatment, and data showed that miR-378d silencing promoted cell survival and increased the number of polyploid tumor cells in TE-1 and KYSE-150 cells (Fig. 6e, P < 0.001) in vitro. Liver-transplantation tumor was used to detect the formation of polyploid tumor cells in vivo, and KYSE-150-NC and KYSE-150-miR-378d-in cells were transferred sequentially. Polyploid tumor cells dramatically increased in liver-transplantation tumor and abdominal-metastasis tumor in the miR-378d silencing group (Fig. 6f). FISH data con rmed that the large cells were polyploid tumor cells (Fig. 6h). These ndings showed that loss of miR-378d promoted polyploid-tumor-cell formation in vitro and in vivo.

miR-378d regulated Hippo-p53 signaling
The downregulation of miR-378d promoted polyploid production in vitro and in vivo. The Hippo-p53 signaling pathway is well known to regulate polyploidy formation, and our GSEA data showed that differentially expressed genes of TE-1 polyploid tumor cells were enriched in the YAP1 and P53 pathway (Fig. 7a). Then, we detected the Hippo-p53 signaling-protein expression levels, and Western blotting data revealed that the transient overexpression of miR-378d mimics increased the expression of p-LATS1/2, p53, and p-YAP. Conversely, the total YAP and TAZ protein expression levels decreased in KYSE510 cells (Fig. 7b), and lost miR-378d expression inhibited the expression of MST1, p-LATS1/2, p53 and p-YAP, thereby increasing the total YAP and TAZ protein expression levels in TE-1 and KYSE510 cells (Fig. 7c).
We also detected the expression of YAP1 and P53 in the cytoplasm and nucleus by Western blotting, and results revealed that YAP1 protein was primarily expressed in the cytoplasm. Moreover, YAP1 protein nuclear translocation increased in KYSE-510 and KYSE-150 cells, and P53 protein nuclear translocation decreased in KYSE-510,TE-1 and KYSE-150 cells (Fig. 7d). Immuno uorescence data indicated that miR-378d absence reduced the TP53 expression and increased the YAP1 and TAZ expression of TE-1 cells in vitro (Fig. 7e). In abdominal-metastasis tumors, loss of miR-378d expression also reduced TP53 expression and increased YAP1 and TAZ expression in In-mice (Fig. 7f, KYSE-150 cells). This nding suggested that lost miR-378d expression inactivated the Hippo-TP53 signaling pathway and activated YAP/TAZ, thereby playing an important role in polyploidy formation in ESCCs.
AKT-β-catenin signaling plays an important role in promoting tumor EMT [35]. and metastasis [36,37]. . We have previously reported that AKT-β-catenin-ALDH1A1 can form a positive feedback regulatory loop to promote the characteristics of cancer stem cells (CSCs) [38,39]. . In the present study, we reported that miR-378d suppressed the AKT-β-catenin circuit loop. AKT and β-catenin also positively regulated each other, as well as the CSC marker ALDH1A1 and the mesenchymal cell marker Vimentin. Furthermore, miR-378d absence promoted EMT and metastasis in vitro and in vivo.
Tumors, even cell lines cultured in vitro, are complex systems that include heterogeneous cancer cells with markedly different sizes and genomic contents [40]. . These cells include bulk, stem, and polyploidy cells. The bulk of cancer cells within the majority are aneuploid. CSCs are often much smaller than bulk cells, whereas polyploid cells are larger owing to their increased ploidy (>4n) [41]. . CD133 has been suggested as a broad-spectrum marker for CSCs. CD133 + cells are small, regular, and round with small microvilli. In some elds, several giant cancer cells the CD133 + cell group have been identi ed under a light microscope, and most of them are found to be polynuclear in NPC cells [42]. . Studies have shown that polyploidy tumor cells express tumor-stem-cell markers. Li Zhang et al. [8]. detected the upregulation of EZH2 expression in polyploid tumor cells and found that a histone lysine N-methyltransferase plays a key role in embryogenesis and CSCs. The expression of ALDH1 was higher in HEY-Polyploid tumor cells than in normal cancer cells, and CD133 and CD117 were expressed in HEY-Polyploid tumor cells but not in normal cancer cells [43]. . Their conclusion is the same as that drawn from a study on PTX-treated breast cancer cell line MCF-7 [44]. , which showed that polyploidy tumor cells are a subgroup of CSCs in normal cancer cells and PTX-treated cells. Other researchers have revealed that polyploid giant cancer cells produce daughter cells with CSC characteristics through budding [43,45]. . Polyploidy tumor cells produce small daughter cells by budding, which may be also be CSCs. Daughter cells are produced by polyploidy tumor cells, which may lead to tumor heterogeneity by differentiation.
In the present study, we identi ed a critical role for miR-378d during cancer-cell polyploidy and in malignant tumor phenotypes. miR-378d expression inhibition increased cancer-cell polyploidy in vitro and in vivo. Cancer-cell polyploidy is important for tumorigenesis and progression, but how polyploid-tumor cells form remain largely unknown. JNK and Yorkie are co-activated to form polyploid tumor cells through endogenous replication, i.e., Yorkie upregulates Diap1 phenotype of and Diap1 and JNK downregulates cyclin B (Cyc B) in G2/M cells, thereby inducing intracellular replication [46]. . Interestingly, malignant tumors induced by RAS activation and cell-polarity defects also comprise polyploid tumor cells, which are also caused by JNK and Yorkie mediated CycB downregulation. Polyploid tumor cells can be eliminated from this malignant tumor though blocking endogenous replication, which strongly inhibits the growth and metastasis of tumors [46]. . Endoduplication cell cycle plays an important role in the development of plants, Drosophila melanogaster, and several mammalian cells and organs including megakaryocytes, hepatocytes, and placentas [47]. . RhoA and AKT play important roles in F-actin skeleton rearrangement [48,49].
. Cytokinesis failure in erythroblasts has been reported in a mouse model with erythroid-speci c deletion of RhoA [50]. .
We found that polyploid-tumor-cell formation by PTX treatment was similar to megakaryocyte formation.
During endomitosis, megakaryocytes proceed through successive cell cycles without cell division to reach DNA contents of 32N, 64N, and even 128N. After completion of the G1, S, and G2 phases, committed megakaryocytes enter mitosis, transition through anaphase, separate their chromosomes, and initiate cleavage-furrow formation [51]. . Polyploid tumor cells treated with PTX undergo endomitosis and separated their chromosomes to their daughter cells through pseudopodia or budding, as in the asymmetric cell-division model [47,52,53]. . In diploid cells, increasing the phosphorylation of LATS1/2 inhibits the transcriptional coactivators YAP and TAZ. In parallel, LATS1/2 stabilizes p53 by disrupting the MDM2-p53 interaction. In polyploid megakaryocytes, the Hippo-p53 pathway remains off as the reduction in RhoA activity fails to activate LATS1/2, allowing YAP/TAZ to translocate into the nucleus and promoting target gene expression [10]. . In the present study, miR-378d inhibition promoted polyploidy in vitro and in vivo, suppressed LATS1/2, p-YAP, and p53 expression, and promoted TAZ protein expression. Moreover, the presence of extra centrosomes and actin laments, a consequence of the increased number of chromosomes, was found to activate the Hippo-p53 pathway by downregulating RhoA activity [54]. . Thus, miR-378d lost expression promotes polyploid cancer cells formed by inactivating the Hippo-p53 signaling pathway and increasing YAP/TAZ nucleus translocation.
PTX stabilizes microtubule polymers, blocking cell cycle in G2-M phase and hindering the development of mitosis. Several studies have shown that PTX promotes polyploid tumor-cell production [43,44,55]. . Thus, considering that polyploid tumor cells promoted chemoresistance and cancer progression, the clinical application of PTX should be re-evaluated or the combination therapy for targeting polyploid should be developed.

Conclusions
We reported for the rst time that low miR-378d expression was associated with poor prognosis in ESCC patients. Low miR-378d expression can suppress Hippo-p53 and promote the AKT-β-catenin signaling pathway, thereby also promoting polyploidy and cancer progression.