This article is mentioned in:

Introduction

Esophageal squamous cell carcinoma (ESCC) is the most prevalent type of esophageal cancer and is the sixth leading cause of cancer mortality worldwide (1). Due to ESCC's aggressive nature, the prognosis of ESCC patients with local invasion and distant metastasis at diagnosis is poor (2,3). Surgical resection is recognized as the preferred treatment for patients with newly diagnosed ESCC. However, high rates of tumor recurrence are notable (4,5). Neoadjuvant chemotherapy or chemoradiotherapy have been demonstrated to prolong overall survival for patients with ESCC (6-8). However, treatment options for recurrent cases are limited, and recently approved targeted therapies have not observed effective therapeutic effects (9,10). Therefore, ESCC patients with recurrence and metastasis require novel and effective treatment strategies.

The latest genomic analyses of ESCC cells have exhibited epigenetic modifications, e.g., DNA methylation, histone deacetylation, chromatin remodeling and non-coding RNA regulation (11,12). In that regard, microRNAs (miRNAs or miRs) consist of a class of small, well-conserved, non-coding RNAs that regulate RNA transcripts in a sequence-dependent manner (13). They participate in physiological and pathological conditions, e.g., cell differentiation, proliferation, motility and metabolism (14). A single miRNA can control a vast number of RNA transcripts in normal and diseased cells (15). Therefore, aberrantly expressed miRNAs may break down regulated RNA networks and contribute to cancer cells' development, metastasis and drug resistance (16).

A large number of miRNAs exhibit differential expression in ESCC, and they contribute to ESCC pathogenesis through their activities as oncogenes or tumor suppressors (11). Analyses of our original miRNA expression signatures by RNA-sequencing revealed that both strands of the miR-145 duplex (miR-145-5p, the guide strand and miR-145-3p, the passenger strand) were significantly downregulated in several types of cancers (17-20). The traditional view of miRNA function has held that only one strand of the miRNA duplex is incorporated into the RNA-induced silencing complex (RISC), becoming the active strand (guide strand). In contrast, the other strand (the passenger strand or miRNA*) was thought to be degraded and to have no function (21,22). However, recent studies of miRNA biogenesis have demonstrated that certain miRNA passenger strands are functional in plant and human cells (20,23).

Our recent studies have demonstrated that both strands of the miR-145 duplex have antitumor roles in lung cancer, bladder cancer, prostate cancer and head and neck cancer (17-20). In ESCC cells, both strands of the miR-150 duplex (miR-150-5p, the guide strand, and miR-150-3p, the passenger strand) acted as antitumor miRNAs through their targeting of SPOCK1 (24). A number of studies demonstrated that miR-145-5p acted as a pivotal antitumor miRNA in human cancers (25), including ESCC (26). In contrast, the functional significance and the targets of miR-145-3p are still obscure.

The aim of the present study was to demonstrate that miR-145-3p possesses antitumor functions and to identify its molecular targets, thereby elucidating ESCC pathogenesis. Thus, in ESCC cells, it was demonstrated that ectopic expression of miR-145-3p significantly blocked cancer cell proliferation, migration and invasion, similar to the actions of miR-145-5p. Furthermore, it was demonstrated that two genes, dehydrogenase/reductase member 2 (DHRS2) and myosin IB (MYO1B), were directly regulated by antitumor miR-145-3p in ESCC cells. Involvement of miR-145-3p (the passenger strand) is a novel concept in ESCC oncogenesis. The present approach, based on the roles of antitumor miRNA and its targets, will contribute to improved understanding of the molecular pathogenesis of ESCC.

Materials and methods

Human ESCC clinical specimens and cell lines

The present study was approved by the Bioethics Committee of Kagoshima University (Kagoshima, Japan; approval no. 28-65). Written prior informed consent and approval were obtained from all of the patients. All subjects in the patient cohort (n=29) were diagnosed with ESCC based upon pathologic criteria. ESCC with curative resection was included, salvage surgery was excluded. From this group, 22 clinical specimens and 12 noncancerous esophageal tissues were obtained. All samples were collected at the Kagoshima University hospital from March 2010 to September 2014. Collection of resected tissues occurred prior to preoperative therapy. The clinicopathological features of the patients are presented in Tables I and II. ESCC is more prevalent among males, thus almost all cases recruited for the present study were male.

Table I Clinicopathological features of esophageal squamous cell carcinoma patients.

Table I

Clinicopathological features of esophageal squamous cell carcinoma patients.

No.	Age (years)	Sex	Differentiation	T	N	M	Stage	ly	v	Recurrence
1	52	Male	Poor	1b	0	0	IA	1	1	+
2	72	Male	Moderate	1b	0	0	IA	0	1	−
3	69	Male	Moderate	1b	0	0	IA	0	0	−
4	56	Male	Moderate	2	0	0	IB	0	1	−
5	66	Male	Moderate	3	0	0	IIA	1	1	−
6	70	Male	Moderate	3	0	0	IIA	1	1	+
7	66	Male	Moderate	3	0	0	IIA	1	1	−
8	71	Male	Well	3	0	0	IIA	1	2	−
9	62	Male	Well	1a	1	0	IIB	0	0	−
10	68	Male	Moderate	1b	1	0	IIB	1	1	−
11	60	Male	Moderate	1b	1	0	IIB	1	1	−
12	71	Male	Moderate	1b	1	0	IIB	0	0	−
13	84	Male	Well	2	1	0	IIB	1	1	−
14	79	Male	Moderate	2	1	0	IIB	1	1	−
15	60	Male	Moderate	2	1	0	IIB	1	2	−
16	68	Male	Poor	1b	2	0	IIIA	1	3	+
17	67	Male	Well	3	2	0	IIIB	2	2	+
18	55	Male	Moderate	3	2	0	IIIB	1	1	+
19	75	Male	Moderate	3	2	0	IIIB	1	1	+
20	74	Male	Moderate	2	3	0	IIIC	3	1	+
21	57	Male	Poor	3	3	0	IIIC	1	1	+
22	63	Male	Well	3	3	0	IIIC	2	1	+

[i] ly, lymphatic invasion; M, metastasis; N, nodes; T, tumor; v, venous invasion.

Table II Features of patients in noncancerous esophageal tissues.

Table II

Features of patients in noncancerous esophageal tissues.

No.	Age (years)	Sex
1	66	Male
2	55	Male
3	52	Male
4	78	Male
5	75	Male
6	60	Male
7	71	Male
8	64	Male
9	79	Female
10	81	Male
11	69	Male
12	84	Male

In addition, 2 ESCC cell lines were used: TE-8, moderately differentiated and TE-9, poorly differentiated (27) (RIKEN BioResource Center, Tsukuba, Japan). Cell culture, extraction of total RNA and extraction of protein were performed as described in previous reports (28,29).

Transfection of mimic and inhibitor miRNA, small interfering (si)RNA into ESCC cells

In the present study, the following mimic and inhibitor miRNAs or siRNAs were transfected: mimic miRNAs (Ambion Pre-miR miRNA precursor; miR-145-5p: 5'-GUCCAGUUUUCCCAGGAAUCCCU-3', ID: PM11480; hsa-miR-145-3p: 5'-GGAUUCCUGGAAAUACUGUUCU-3', ID: PM13036; Thermo Fisher Scientific, Inc., Waltham, MA, USA), inhibitor miRNAs (Anti-miR miRNA Inhibitor; has-miR-145-3p: 5'-GGAUUCCUGGAAAUACUG UUCU-3', ID: AM13036; Applied Biosystems; Thermo Fisher Scientific, Inc.) and siRNAs (Stealth Select RNAi siRNA; si-DHRS2, ID: HSS145497 and HSS173461; si-MYO1B, ID: HSS106714 and HSS106716; Invitrogen; Thermo Fisher Scientific, Inc.) and negative control miRNA/siRNA (product ID: AM17111; Thermo Fisher Scientific, Inc.). The transfection procedures were performed as previously described (28-30).

Incorporation of miR-145-3p into the RISC: Assessment by argonaute 2 (Ago2) immunoprecipitation

miRNAs were transfected into TE-8 cells and miRNAs were isolated using an microRNA Isolation Kit, Human Ago2 (Wako Pure Chemical Industries, Ltd., Osaka, Japan) as described previously (28-30). The expression levels of Ago2-conjugated miRNAs were assessed by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). miR-21 (assay ID; 000397; Applied Biosystems) was used as the internal control.

RT-qPCR

Quantification of miRNAs and mRNAs was performed by StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific, Inc.). The procedure used for RT-qPCR has been described previously (28,29). The expression levels of miRNAs were analyzed using TaqMan RT-qPCR assays (miR-145-5p, assay ID: 002278; miR-145-3p assay ID: 002149; Applied Biosystems; Thermo Fisher Scientific, Inc.). Data were normalized to RNU48 (assay ID: 001006; Applied Biosystems). In addition, the expression levels of DHRS2 and MYO1B were assessed with the following TaqMan probes: DHRS2, assay ID: Hs01061575_g1; MYO1B, assay ID: Hs00362654_m1; Applied Biosystems; Thermo Fisher Scientific, Inc.), and normalized to glucuronidase β (assay ID: Hs00939627_ml; Applied Biosystems; Thermo Fisher Scientific, Inc.).

Cell proliferation, migration, invasion and apoptosis assays

Protocols for determining cell proliferation (XTT assays), migration and invasion were described previously (28,29). For apoptosis assays, double staining with fluorescein isothiocyanate (FITC)-Annexin V and propidium iodide was carried out using a FITC Annexin V Apoptosis Detection kit (BD Biosciences, Franklin Lakes, NJ, USA) according to the manufacturer's recommendations. Stains were analyzed within 1 h using a flow cytometer (CyAn ADP analyzer; Beckman Coulter, Inc., Brea, CA, USA). Cells were identified as viable cells, dead cells, early apoptotic cells and late apoptotic cells using Summit 4.3 software (Beckman Coulter, Inc.). The percentages of early apoptotic and late apoptotic cells from each experiment were then compared. As a positive control, 1 µM gemcitabine hydrochloride (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan) was used.

Identification of putative target genes regulated by miR-145-3p in ESCC cells

The present strategy for identification of miR-145-3p target genes is outlined in Fig. 1. The microarray data were deposited in the Gene Expression Omnibus (GEO) repository (https://www.ncbi.nlm.nih.gov/geo/) under accession number GSE107008. Putative target genes with a binding site for miR-145-3p were detected by TargetScanHuman ver.7.1 (http://www.targetscan.org/vert_71/). The GEO database (GSE20347) was used for assessment of the association between target genes and ESCC.

Figure 1 Strategy for identification of putative target genes regulated by miR-145-3p in ESCC cells. Gene expression analyses demonstrated that 1,374 genes were downregulated in miR-145-3p-transfected TE-8 cells. Finally, 30 genes were selected as putative targets of miR-145-3p in ESCC cells. miR, microRNA; ESCC, esophageal squamous cell carcinoma.

Exploration of downstream targets regulated by si-DHRS2 and si-MYO1B in ESCC

Genome-wide microarray analysis was used for identification of DHRS2 and MYO1B downstream targets. Expression data were deposited in a GEO database (GSE118966). A GEO database (GSE20347) was used for assessment of the association between target genes and ESCC. Our strategy for identification of DHRS2 and MYO1B downstream targets is outlined in Fig. 2. Expression analysis was performed on microarray data using SurePrint G3 Human 8×60K v3 (Agilent Technologies, Inc., Santa Clara, CA, USA).

Figure 2 Strategy for identification of putative downstream genes regulated by (A) si-DHRS2 and (B) si-MYO1B in TE-8 cells. Microarray analyses were performed using TE-8 cells in which DHRS2 or MYO1B were knocked down to identify candidate genes that were downstream. si, small interfering RNA; DHRS2, dehydrogenase/reductase member 2; MYO1B, myosin IB.

Western blot analysis

Anti-human DHRS2 rabbit polyclonal immunoglobulin (Ig)G (1:1,000; HPA069551; Sigma-Aldrich; Merck KGaA) and anti-human MYO1B rabbit polyclonal IgG (1:250; HPA013607; Sigma-Aldrich; Merck KGaA) were used as primary antibodies. Anti-human β-actin mouse monoclonal IgG (1:2,000; A1978; Sigma-Aldrich; Merck KGaA) was used as an internal control. The protocol for Western blot analysis was described previously (29,30).

Immunohistochemistry

Tumor specimens were fixed, embedded and sectioned as described previously (31). Anti-human DHRS2 rabbit polyclonal IgG (1:250; HPA069551; Sigma-Aldrich, St. Louis, MO, USA) and anti-human MYO1B rabbit polyclonal IgG (1:300; HPA013607; Sigma-Aldrich; Merck KGaA were used as primary antibodies. The protocol followed was described previously (32).

Luciferase reporter assays

The following sequences were inserted into the psiCHECk-2 vector (C8021; Promega Corporation, Madison, WI, USA): The wild-type sequences of the 3'-untranslated regions (UTRs) of DHRS2 and MYO1B, or the deletion-type, which lacks the miR-145-3p target sites from DHRS2 (position 270-276) or MYO1B (position 88-94 or position 1,117-1,123). The cloned vectors were co-transfected into ESCC cells with mature miR-145-3p. The procedures for transfection and dual-luciferase reporter assays have been reported previously (28,29).

Statistical analysis

Associations between groups were analyzed using the Mann-Whitney U test or Tukey's multiple comparisons test following one-way analysis of variance. The differences between survival rates were analyzed by Kaplan-Meier survival curves and log-rank statistics. Spearman's rank test was used to evaluate the correlations between the expression levels of miR-145-3p, miR-145-5p, DHRS2 and MYO1B. Data are presented as the mean ± standard deviation. P<0.05 was considered to indicate a statistically significant difference. Expert StatView version 5.0 (SAS Institute, Inc., Cary, NC, USA) and GraphPad Prism version 7.04 (GraphPad Software, Inc., La Jolla, CA, USA) were used in these analyses.

Results

Expression levels of miR-145-5p and miR-145-3p in ESCC clinical specimens

Expression levels of miR-145-5p and miR-145-3p were significantly downregulated in cancer tissues and ESCC cell lines relative to normal tissues (Fig. 3A). Spearman's rank test demonstrated a positive correlation between the expression levels of miR-145-5p and miR-145-3p (Fig. 3B).

Figure 3 Effects of ectopic expression of miR-145-5p and miR-145-3p on ESCC cells. (A) Expression levels of miR-145-5p and miR-145-3p in ESCC clinical specimens and cell lines. RNU48 was used as an internal control. (B) Spearman's rank test demonstrated a positive correlation between the expression levels of miR-145-5p and miR-145-3p. (C) The 5-year survival rates were analyzed by Kaplan-Meier survival curves and log-rank statistics. (D) Expression levels of miR-145-5p and miR-143-3p in ESCC cell lines. RNU48 was used as an internal control. TE-8 and TE-9 cells were transfected for 72 h following transfection with 10 nM miR-145-5p and miR-143-3p mimic. (E) Cell proliferation was determined by XTT assays 72 h following transfection with miR-145-5p and miR-145-3p. (F) Results of cell migration assays. (G) Cell invasion activity was determined using Matrigel invasion assays. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. miR, microRNA; ESCC, esophageal squamous cell carcinoma.

It was demonstrated that 5-year survival was significantly higher in patients who had high miR-145-3p expression than in patients with low miR-145-3p expression (Fig. 3C). There was no significant association between the expression level of miR-145-5p and patient survivals (Fig. 3C).

Ectopic expression of miR-145-5p and miR-145-3p: Impact on ESCC cells

To assess the ectopic expression of miR-145-5p and miR-145-3p, the mimic miRNA was transfected to ESCC cell lines (Fig. 3D). XTT assays demonstrated significant inhibition of cell proliferation in miR-145-5p and miR-145-3p transfectants (Fig. 3E). Likewise, cell migration and invasion were significantly inhibited following miR-145-5p or miR-145-3p transfection (Fig. 3F and G). The numbers of early apoptotic and late apoptotic cells were significantly larger in miR-145-5p or miR-145-3p transfectants than in mock or negative control transfectants (Fig. 4).

Figure 4 Effects of miR-145-5p and miR-145-3p on TE-8 cell apoptosis. Apoptosis assays were performed using a fluorescein isothiocyanate Annexin V apoptosis detection kit with flow cytometric determination. Early apoptotic cells are in area R2 and late apoptotic cells are in area R3. The normalized ratios of early and late apoptotic cells are presented in the histogram. GEM was used as a positive control. *P<0.05, ****P<0.0001. miR, microRNA; GEM, gemcitabine hydrochloride.

Incorporation of miR-145-3p into the RISC in ESCC cells

It was anticipated that the passenger strand of miR-145-3p was incorporated into the RISC and served as a tumor suppressor in ESCC cells. To verify that hypothesis, Ago2 was immunoprecipitated in cells that had been transfected with either miR-145-5p or miR-145-3p (Fig. 5A). Ago2 is an essential component of the RISC (16). Isolated Ago2-bound miRNAs were analyzed by RT-qPCR to confirm whether miR-145-5p and miR-145-3p bound to Ago2. In TE-8 cells, miR-145-5p transfectants demonstrated higher expression levels of miR-145-5p than mock transfectants, miR-control or miR-145-3p transfectants. Similarly, following miR-145-3p transfection, miR-145-3p was detected by Ago2 immunoprecipitation (Fig. 5B). miR-145-5p and miR-145-3p were demonstrated to bind to Ago2 separately and were incorporated into RISC, thereby demonstrating miRNA function.

Figure 5 Both strands of miR-145 duplex bound to Ago2. (A) Schematic illustration of miRNA detection methods. Isolation of RISC incorporated miRNAs by Ago2 immunoprecipitation. (B) Expression levels of miRNAs bound to Ago2 were assessed by reverse transcription-quantitative polymerase chain reaction. Expression data were normalized by the expression of miR-21. ****P<0.0001. miR/miRNA, microRNA; Ago2, argonaute 2; RISC, RNA-induced silencing complex.

Effects of co-transfection of mimic and inhibitor miR-145-3p into ESCC cells

To confirm the antitumor effects of miR-145-3p, rescue experiments were performed using mimic and inhibitor miR-145-3p with TE-8 cells (Fig. 6A). The rescue experiments indicated that cancer cell proliferation, migration and invasion were rescued in miR-145-3p inhibitor transfectants compared with restored miR-145-3p mimic only (Fig. 6B-D).

Figure 6 Rescue experiments: Tumor suppressive effect of miR-145-3p by mimic and inhibitor using ESCC cells. TE-8 cells were transfected with 10 nM miR-145-3p mimic to overexpress miR-145-3p; the tumor suppressive effects of TE-8 were readily observed. However, co-transfection of 10 nM miR-145-3p mimic and inhibitor demonstrated that the tumor suppressive effects of miR-145-3p were attenuated. (A) Expression levels of miR-145-3p in ESCC cell lines. RNU48 was used as an internal control. TE-8 cells were transfected for 72 h following co-transfection with 10 nM miR-145-3p mimic and inhibitor. (B) Cell proliferation activity was assessed by XTT assays. TE-8 cells were transfected for 72 h following co-transfection with 10 nM miR-145-3p mimic and inhibitor. (C) Cell migration activity was assessed 72 h following co-transfection with 10 nM miR-145-3p mimic and inhibitor. (D) Cell invasion activity was determined using Matrigel 72 h following co-transfection with 10 nM miR-145-3p mimic and inhibitor. *P<0.05, **P<0.01, ****P<0.0001. miR, microRNA; ESCC, esophageal squamous cell carcinoma.

Searching for putative targets regulated by miR-145-3p in ESCC cells

The strategy to identify miR-145-3p target genes is presented in Fig. 1. Gene expression analyses demonstrated that 1,374 genes were downregulated (log2 ratio <-1.0) in miR-145-3p transfected TE-8 cells compared with control transfectants. The present expression data were deposited in the GEO repository under accession no. GSE107008. Among these downregulated genes, genes that had putative miR-145-3p binding sites in their 3'-UTRs were selected using information in the TargetScan database. A total of 280 genes were identified. Then, 30 genes were selected by restricting the identified genes to those strongly upregulated in ESCC clinical specimens (log2 ratio >1.0; GEO accession no. GSE20347; Table III).

Table III Putative targets of miR-145-3p regulation in ESCC cells.

Table III

Putative targets of miR-145-3p regulation in ESCC cells.

Entrez gene ID	Gene symbol	Gene name	TE-8 miR-145-3p transfectant	ESCC GSE20347 fold-change	Target site count	Prognosis P-value: TCGA OncoLnc data
						ESCA	ESCC
10202	DHRS2	Dehydrogenase/reductase (SDR family) member 2	−2.69	2.02	1	0.047	0.708
1848	DUSP6	Dual specificity phosphatase 6	−2.61	1.00	1	0.456	0.469
55157	DARS2	Aspartyl-tRNA synthetase 2, mitochondrial	−2.09	1.17	2	0.504	0.706
6646	SOAT1	Sterol O-acyltransferase 1	−2.06	1.81	1	0.732	0.667
4430	MYO1B	Myosin IB	−1.98	1.61	2	0.372	0.856
2115	ETV1	Ets variant 1	−1.84	1.10	1	0.142	0.119
983	CDK1	Cyclin-dependent kinase 1	−1.63	1.95	1	0.621	0.136
1719	DHFR	Dihydrofolate reductase	−1.48	1.14	1	0.199	0.465
51053	GMNN	Geminin, DNA replication inhibitor	−1.46	1.37	1	0.274	0.189
23321	TRIM2	Tripartite motif containing 2	−1.35	1.45	1	<0.001a	0.037a
55697	VAC14	Vac14 homolog (S. cerevisiae)	−1.34	1.53	1	0.052	0.095
79789	CLMN	Calmin (calponin-like, transmembrane)	−1.33	1.79	2	0.519	0.352
5654	HTRA1	HtrA serine peptidase 1	−1.30	1.44	1	0.863	0.252
54830	NUP62CL	Nucleoporin 62 kDa C-terminal like	−1.28	1.10	1	0.313	0.313
204	AK2	Adenylate kinase 2	−1.24	1.21	2	0.691	0.972
126321	MFSD12	Major facilitator superfamily domain containing 12	−1.23	1.05	1	0.838	0.098
9532	BAG2	BCL2-associated athanogene 2	−1.20	1.80	2	0.109	0.563
51029	DESI2	Desumoylating isopeptidase 2	−1.18	1.25	2	0.261	0.19
6711	SPTBN1	Spectrin, β, non-erythrocytic 1	−1.18	1.21	1	0.515	0.504
1163	CKS1B	CDC28 protein kinase regulatory subunit 1B	−1.14	2.02	1	0.658	0.658
8534	CHST1	Carbohydrate (keratan sulfate Gal-6) Sulfotransferase 1	−1.14	1.52	1	0.943	0.983
5174	PDZK1	PDZ domain containing 1	−1.14	1.65	1	0.565	0.462
875	CBS	Cystathionine-β-synthase	−1.14	3.55	2	0.199	0.160
23516	SLC39A14	Solute carrier family 39 (zinc transporter), member 14	−1.12	2.68	1	0.322	0.101
6790	AURKA	Aurora kinase A	−1.07	2.17	1	0.322	0.051
79718	TBL1XR1	Transducin (β)-like 1 X-linked receptor 1	−1.06	1.30	1	0.579	0.377
23141	ANKLE2	Ankyrin repeat and LEM domain containing 2	−1.05	1.16	1	0.418	0.585
23649	POLA2	Polymerase (DNA directed), α 2, accessory subunit	−1.04	1.35	1	0.642	0.842
64151	NCAPG	Non-SMC condensin I complex, subunit G	−1.03	1.22	1	0.198	0.056
22848	AAK1	AP2 associated kinase 1	−1.03	1.27	3	0.634	0.346

a Poor prognosis with low expression. miR, microRNA; ESCC, esophageal squamous cell carcinoma; ESCA, esophageal carcinoma.

In this fashion, DHRS2 was focused on, as its expression was the most downregulated in miR-145-3p transfectants and the most upregulated in ESCC clinical specimens. Additionally, MYO1B was examined, as it was more highly downregulated in miR-145-3p transfectants and was more upregulated in ESCC clinical specimens. In addition, our previous study had demonstrated that the activation of MYO1B was associated with cancer cell aggressiveness (20).

Direct regulation of DHRS2 and MYO1B by miR-145-3p in ESCC cells

The finding that DHRS2 and MYO1B were downregulated by expression of miR-145-3p was further investigated (Figs. 7 and 8). ESCC cells (TE-8) that had been transfected with miR-145-3p were examined. Using RT-qPCR, it was demonstrated that DHRS2 and MYO1B mRNA levels were significantly reduced by miR-145-3p transfection (Figs. 7A and 8A). Furthermore, western blot analysis was performed to measure the expression levels of DHRS2 and MYO1B proteins in the transfectants. Results demonstrated that the proteins were also reduced by miR-145-3p transfection (Figs. 7B and 8B).

Figure 7 Direct regulation of DHRS2 by miR-145-3p in esophageal squamous cell carcinoma cells. (A) Expression levels of DHRS2 mRNA 72 h following transfection with 10 nM miR-145-3p into cell lines. Glucuronidase β was used as an internal control. (B) Expression of DHRS2 proteins 72 h following transfection with miR-145-3p. β-actin was the loading control. (C) miR-145-3p binding site (positions 270-276) in the 3'-UTR of DHRS2 mRNA. (D) Dual luciferase reporter assays using vectors encoding putative miR-145-3p target site in the DHRS2 3'-UTRs for wild-type and deleted regions. Renilla luciferase values were normalized to firefly luciferase values. ****P<0.0001. DHRS2, dehydrogenase/reductase member 2; miR, microRNA; UTR, untranslated region.

Figure 8 Direct regulation of MYO1B by miR-145-3p in esophageal squamous cell carcinoma cells. (A) Expression levels of MYO1B mRNA 72 h following transfection with 10 nM miR-145-3p into cell lines. Glucuronidase β was used as an internal control. (B) Expression of MYO1B proteins 72 h following transfection with miR-145-3p. β-actin was the loading control. (C) miR-145-3p binding sites (positions 88-94 or positions 1,117-1,123) in the 3'-UTR of MYO1B mRNA. (D) Dual luciferase reporter assays using vectors encoding putative miR-145-3p target sites in the MYO1B 3'-UTRs for both wild-type and deleted regions. Renilla luciferase values were normalized to firefly luciferase values. ****P<0.0001. MYO1B, myosin IB; miR, microRNA.

Whether miR-145-3p directly regulated DHRS2 and MYO1B genes in a sequence-dependent manner was then evaluated.

The Human TargetScan database predicted that DHRS2 had 1 binding site (positions 270-276) for miR-145-3p in the 3'-UTR (Fig. 7C). Accordingly, luciferase reporter assays were carried out with vectors that included either the wild-type or deletion-type 3'-UTR of DHRS2. Co-transfection with miR-145-3p and vectors including the wild-type sequence significantly reduced luciferase activity compared with those in mock and miR-control transfectants in position 270-276 of the DHRS2 3'-UTR (Fig. 7D).

MYO1B had 2 binding sites (positions 88-94 and 1,117-1,123) for miR-145-3p in the 3'-UTR (Fig. 8C). Luciferase activities were significantly reduced in position 1,117-1,123 of the MYO1B 3'-UTR (Fig. 8D).

Effects of silencing DHRS2 and MYO1B in ESCC cells

Subsequently, siRNAs were transfected into TE-8 cells to examine the function of DHRS2 and MYO1B in ESCC cells (Figs. 9 and 10). The mRNA and protein expression levels of DHRS2 and MYO1B were decreased by si-DHRS2 and si-MYO1B, respectively (Figs. 9A and B, and 10A and B). Subsequently, the effects of DHRS2 or MYO1B knockdown on ESCC cell proliferation, migration and invasion were investigated.

Figure 9 Effects of silencing DHRS2 in ESCC cells. (A) DHRS2 mRNA expression 72 h following transfection of 10 nM si-DHRS2 into ESCC cell lines. Glucuronidase β was used as an internal control. (B) Protein expression 72 h following transfection with si-DHRS2. β-actin was the loading control. (C) Cell proliferation was determined with XTT assays 72 h following transfection with 10 nM si-DHRS2. (D) Results of cell migration assays. (E) Results of Matrigel invasion assays. (F) Results of cell apoptosis assays. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. DHRS2, dehydrogenase/reductase member 2; ESCC, esophageal squamous cell carcinoma; si, small interfering RNA.

Figure 10 Effects of silencing MYO1B in ESCC cells. (A) MYO1B mRNA expression 72 h following transfection of 10 nM si-MYO1B into ESCC cell lines. Glucuronidase β was used as an internal control. (B) Protein expression 72 h following transfection with si-MYO1B. β-actin was the loading control. (C) Cell proliferation was determined with XTT assays 72 h following transfection with 10 nM si-MYO1B. (D) Results of cell migration assays. (E) Results of Matrigel invasion assays. (F) Results of cell apoptosis assays. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. MYO1B, myosin IB; ESCC, esophageal squamous cell carcinoma; si, small interfering RNA.

Cancer cell proliferation was significantly suppressed in si-DHRS2 or si-MYO1B transfectants compared with mock and si-RNA-control transfectants. Additionally, migration and invasion activities were significantly inhibited in si-DHRS2 or si-MYO1B transfectants (Figs. 9C-E and 10C-E). In the apoptosis assays, si-DHRS2_1/si-DHRS2_2 and si-MYO1B_1/si-MYO1B_2 transfections significantly increased apoptotic TE-8 cells (Figs. 9F and 10F).

Expression of DHRS2 and MYO1B in ESCC clinical specimens

Based upon the findings above, it was of great interest to use RT-qPCR to determine the expression levels of DHRS2 and MYO1B in clinical specimens. DHRS2 and MYO1B expression levels were significantly upregulated in ESCC tumor tissues (Fig. 11A). Additionally, the 5-year survival rates of ESSC patients were significantly shorter in those with elevated DHRS2 expression compared with those with low expression (Fig. 11B). There was no significant association between the expression levels of MYO1B and the survival rate (Fig. 11B).

Figure 11 Expression of DHRS2 and MYO1B in ESCC clinical specimens. (A) Expression levels of DHRS2 and MYO1B in ESCC clinical specimens. GUSB was the internal control. (B) The 5-year survival rates were analyzed by Kaplan-Meier survival curves and log-rank statistics. (C) Spearman's rank test was used to evaluate the correlation between DHRS2 or MYO1B expression and miR-145-3p. DHRS2, dehydrogenase/reductase member 2; MYO1B, myosin IB; ESCC, esophageal squamous cell carcinoma; miR, microRNA; GUSB, glucuronidase β.

Spearman's rank test demonstrated a negative correlation between the expression of DHRS2 and miR-145-3p, and MYO1B and miR-145-3p (Fig. 11C). Furthermore, the protein expression levels of DHRS2 and MYO1B were examined in ESCC clinical specimens by immunostaining. Both DHRS2 and MYO1B were strongly expressed in cancer tissues, but not in noncancerous epithelia (Fig. 12).

Figure 12 Immunohistochemical analysis of esophageal squamous cell carcinoma clinical samples. Immunostaining demonstrated that DHRS2 and MYO1B were strongly expressed in cancer tissues whereas strong expression was not observed in non-cancerous epithelia (x100 and x400 magnification fields). DHRS2, dehydrogenase/reductase member 2; MYO1B, myosin IB; H&E, hematoxylin and eosin.

Exploration of downstream targets regulated by si-DHRS2 and si-MYO1B in ESCC

The present strategy for selecting downstream genes regulated by DHRS2 and MYO1B is demonstrated in Fig. 2. A total of 756 genes were commonly downregulated (log2 ratio <-1.5) in si-DHRS2-transfected TE-8 cells. The upregulated genes in ESCC tissues were also assessed by GEO database analyses (GEO accession no. GSE20347). With that approach, 85 candidate genes downstream from DHRS2 were identified (Table IV). Furthermore, a total of 334 genes were commonly downregulated (log2 ratio <−1.5) in si-MYO1B-transfected TE-8 cells. In a similar approach, 32 candidate genes downstream from MYO1B were identified (Table V).

Table IV Putative targets of si-DHRS2 regulation in ESCC cells.

Table IV

Putative targets of si-DHRS2 regulation in ESCC cells.

Entrez gene ID	Gene symbol	Gene name	ESCC GSE20347 fold-change	TE-8 si-DHRS2 transfectant
1592	CYP26A1	Cytochrome P450, family 26, subfamily A, polypeptide 1	1.86	−3.78
10112	KIF20A	Kinesin family member 20A	1.50	−3.28
259266	ASPM	Asp (abnormal spindle) homolog, microcephaly associated (Drosophila)	1.56	−3.08
27074	LAMP3	Lysosomal-associated membrane protein 3	1.56	−3.03
11098	PRSS23	Protease, serine, 23	1.86	−3.02
10202	DHRS2	Dehydrogenase/reductase (SDR family) member 2	2.02	−2.98
4322	MMP13	Matrix metallopeptidase 13 (collagenase 3)	5.12	−2.93
79075	DSCC1	DNA replication and sister chromatid cohesion 1	1.95	−2.80
983	CDK1	Cyclin-dependent kinase 1	1.95	−2.79
332	BIRC5	Baculoviral IAP repeat containing 5	1.59	−2.72
995	CDC25C	Cell division cycle 25C	1.51	−2.71
11065	UBE2C	Ubiquitin-conjugating enzyme E2C	1.68	−2.70
4751	NEK2	NIMA-related kinase 2	1.66	−2.70
1033	CDKN3	Cyclin-dependent kinase inhibitor 3	1.94	−2.70
11339	OIP5	Opa interacting protein 5	1.63	−2.68
2842	GPR19	G protein-coupled receptor 19	2.12	−2.64
10615	SPAG5	Sperm associated antigen 5	1.59	−2.60
9055	PRC1	Protein regulator of cytokinesis 1	1.58	−2.60
699	BUB1	BUB1 mitotic checkpoint serine/threonine kinase	2.04	−2.59
991	CDC20	Cell division cycle 20	1.54	−2.58
55355	HJURP	Holliday junction recognition protein	1.79	−2.51
7153	TOP2A	Topoisomerase (DNA) II alpha 170 kDa	1.91	−2.46
10403	NDC80	NDC80 kinetochore complex component	1.76	−2.45
55388	MCM10	Minichromosome maintenance complex component 10	1.90	−2.45
2263	FGFR2	Fibroblast growth factor receptor 2	1.65	−2.44
3161	HMMR	Hyaluronan-mediated motility receptor (RHAMM)	1.60	−2.42
1063	CENPF	Centromere protein F, 350/400 kDa	2.31	−2.40
7272	TTK	TTK protein kinase	1.58	−2.36
9401	RECQL4	RecQ protein-like 4	1.92	−2.35
9355	LHX2	LIM homeobox 2	2.63	−2.33
22836	RHOBTB3	Rho-related BTB domain containing 3	1.98	−2.33
54478	FAM64A	Family with sequence similarity 64, member A	1.60	−2.33
9133	CCNB2	Cyclin B2	1.68	−2.27
9787	DLGAP5	Discs, large (Drosophila) homolog-associated protein 5	1.72	−2.26
9156	EXO1	Exonuclease 1	1.89	−2.26
3833	KIFC1	Kinesin family member C1	2.15	−2.19
347733	TUBB2B	Tubulin, beta 2B class IIb	1.86	−2.18
220134	SKA1	Spindle and kinetochore associated complex subunit 1	1.73	−2.18
4291	MLF1	Myeloid leukemia factor 1	1.90	−2.15
8438	RAD54L	RAD54-like (S. cerevisiae)	2.31	−2.14
79019	CENPM	Centromere protein M	2.10	−2.14
51514	DTL	Denticleless E3 ubiquitin protein ligase homolog (Drosophila)	1.62	−2.12
55872	PBK	PDZ binding kinase	1.70	−2.11
3790	KCNS3	Potassium voltage-gated channel, modifier subfamily S, member 3	2.32	−2.11
51512	GTSE1	G-2 and S-phase expressed 1	2.02	−2.10
9493	KIF23	Kinesin family member 23	1.96	−2.10
5983	RFC3	Replication factor C (activator 1) 3, 38 kDa	1.74	−2.09
81611	ANP32E	Acidic (leucine-rich) nuclear phosphoprotein 32 family, member E	1.52	−2.07
83461	CDCA3	Cell division cycle associated 3	2.14	−2.05
23350	U2SURP	U2 snRNP-associated SURP domain containing	1.62	−2.01
6790	AURKA	Aurora kinase A	2.17	−2.00
55165	CEP55	Centrosomal protein 55 kDa	1.94	−1.99
80178	C16orf59	Chromosome 16 open reading frame 59	1.61	−1.98
2305	FOXM1	Forkhead box M1	2.16	−1.96
24137	KIF4A	Kinesin family member 4A	1.95	−1.94
22974	TPX2	TPX2, microtubule-associated	1.65	−1.94
55215	FANCI	Fanconi anemia, complementation group I	1.70	−1.91
10635	RAD51AP1	RAD51 associated protein 1	2.20	−1.88
993	CDC25A	Cell division cycle 25A	1.88	−1.88
2175	FANCA	Fanconi anemia, complementation group A	1.93	−1.85
4171	MCM2	Minichromosome maintenance complex component 2	2.55	−1.82
2491	CENPI	Centromere protein I	1.81	−1.81
655	BMP7	Bone morphogenetic protein 7	1.54	−1.77
4998	ORC1	Origin recognition complex, subunit 1	1.53	−1.76
10036	CHAF1A	Chromatin assembly factor 1, subunit A (p150)	1.75	−1.76
4085	MAD2L1	MAD2 mitotic arrest deficient-like 1 (yeast)	1.67	−1.76
3149	HMGB3	High mobility group box 3	2.01	−1.74
29028	ATAD2	ATPase family, AAA domain containing 2	1.96	−1.73
9837	GINS1	GINS complex subunit 1 (Psf1 homolog)	1.64	−1.73
51659	GINS2	GINS complex subunit 2 (Psf2 homolog)	1.86	−1.72
5984	RFC4	Replication factor C (activator 1) 4, 37 kDa	2.08	−1.69
55839	CENPN	Centromere protein N	1.72	−1.66
7078	TIMP3	TIMP metallopeptidase inhibitor 3	2.23	−1.66
27346	TMEM97	Transmembrane protein 97	1.67	−1.65
9928	KIF14	Kinesin family member 14	2.14	−1.64
3625	INHBB	Inhibin, β B	1.75	−1.63
10721	POLQ	Polymerase (DNA directed), θ	1.51 -	1.62
1663	DDX11	DEAD/H (Asp-Glu-Ala-Asp/His) box helicase 11	2.08	−1.60
9319	TRIP13	Thyroid hormone receptor interactor 13	2.02	−1.58
4605	MYBL2	V-myb avian myeloblastosis viral oncogene	3.08	−1.56
		homolog-like 2
51762	RAB8B	RAB8B, member RAS oncogene family	1.76	−1.54
91860	CALML4	Calmodulin-like 4	1.90	−1.54
10293	TRAIP	TRAF interacting protein	1.50	−1.52
2237	FEN1	Flap structure-specific endonuclease 1	1.54	−1.51
55753	OGDHL	Oxoglutarate dehydrogenase-like	2.37	−1.51

[i] DHRS2, dehydrogenase/reductase member 2; ESCC, esophageal squamous cell carcinoma; si, small interfering RNA.

Table V Putative targets of si-MYO1B regulation in ESCC cells.

Table V

Putative targets of si-MYO1B regulation in ESCC cells.

Entrez gene ID	Gene symbol	Gene name	ESCC GSE20347 fold-change	TE-8 si-MYO1B transfectant
4430	MYO1B	Myosin IB	1.61	−3.60
347733	TUBB2B	Tubulin, β 2B class IIb	1.86	−2.89
9837	GINS1	GINS complex subunit 1 (Psf1 homolog)	1.64	−2.72
4322	MMP13	Matrix metallopeptidase 13 (collagenase 3)	5.12	−2.66
79075	DSCC1	DNA replication and sister chromatid cohesion 1	1.95	−2.55
4312	MMP1	Matrix metallopeptidase 1 (interstitial collagenase)	6.54	−2.40
51514	DTL	Denticleless E3 ubiquitin protein ligase homolog (Drosophila)	1.62	−2.38
4319	MMP10	Matrix metallopeptidase 10 (stromelysin 2)	4.51	−2.28
23657	SLC7A11	Solute carrier family 7 (anionic amino acid transporter light chain, xc- system), member 11	1.97	−2.20
5983	RFC3	Replication factor C (activator 1) 3, 38 kDa	1.74	−2.11
10202	DHRS2	Dehydrogenase/reductase (SDR family) member 2	2.02	−2.04
2491	CENPI	Centromere protein I	1.81	−2.01
4085	MAD2L1	MAD2 mitotic arrest deficient-like 1 (yeast)	1.67	−1.99
6574	SLC20A1	Solute carrier family 20 (phosphate transporter), member 1	1.52	−1.91
4998	ORC1	Origin recognition complex, subunit 1	1.53	−1.84
81611	ANP32E	Acidic (leucine-rich) nuclear	1.52	−1.83
		phosphoprotein 32 family, member E
10669	CGREF1	Cell growth regulator with EF-hand domain 1	1.62	−1.81
55388	MCM10	Minichromosome maintenance complex component 10	1.90	−1.79
2119	ETV5	Ets variant 5	2.05	−1.79
11199	ANXA10	Annexin A10	2.06	−1.76
655	BMP7	Bone morphogenetic protein 7	1.54	−1.73
983	CDK1	Cyclin-dependent kinase 1	1.95	−1.72
55872	PBK	PDZ binding kinase	1.70	−1.70
4072	EPCAM	Epithelial cell adhesion molecule	2.56	−1.68
8914	TIMELESS	Timeless circadian clock	1.54	−1.65
9518	GDF15	Growth differentiation factor 15	1.58	−1.65
55215	FANCI	Fanconi anemia, complementation group I	1.70	−1.62
11339	OIP5	Opa interacting protein 5	1.63	−1.61
10635	RAD51AP1	RAD51 associated protein 1	2.20	−1.59
332	BIRC5	Baculoviral IAP repeat containing 5	1.59	−1.58
995	CDC25C	Cell division cycle 25C	1.51	−1.54
8438	RAD54L	RAD54-like (S. cerevisiae)	2.31	−1.52

[i] MYO1B, myosin IB; ESCC, esophageal squamous cell carcinoma; si, small interfering RNA.

Discussion

Downregulation of miR-145-5p has frequently been observed in a wide range of cancers, including ESCC (32). A number of previous studies have demonstrated that ectopic expression of miR-145-5p suppressed cancer cell proliferation, migration, invasion and drug resistance both in vitro and in vivo (25,26,33). Notably, the promoter region of pre-miR-145 has a p53 response element and its expression is controlled by activation of p53 under various conditions (34). Therefore, both strands of the miR-145 duplex are pivotal tumor suppressor miRNAs controlled by p53.

Analyses of the miRNA expression signatures demonstrated that certain miRNA passenger strands were downregulated and acted as antitumor miRNAs in several cancers, e.g., miR-145-3p, miR-150-3p, miR-148a-5p and miR-99a-3p (20,24,35,36). Our previous studies demonstrated that antitumor miR-145-3p directly targeted oncogenes, e.g., MTDH in lung adenocarcinoma, UHRF1 in bladder cancer, MYO1B in head and neck cancer and MELK, NCAPG, BUB1 and CDK1 in prostate cancer (17-20). Another group demonstrated that miR-145-3p inhibited cell growth, motility and chemotaxis in non-small cell lung cancer by targeting pyruvate dehydrogenase kinase 1 through suppressing the mechanistic target of rapamycin pathway (37). These findings indicated that antitumor miR-145-3p is associated with cancer pathogenesis.

Exploring the RNA network controlled by antitumor miR-145-3p expands the understanding of the novel molecular pathogenesis of ESCC. In the present study, 30 genes were identified as putative oncogenes based on miR-145-3p regulation in ESCC cells. Among these genes, the following 3 were reported to be cancer-promoting genes in ESCC: cyclin dependent kinase 1 (CDK1), aurora kinase A (AURKA) and transducin β like 1 X-linked receptor 1 (TBL1XR1) (38-43). Identification of the target genes controlled by miR-145-3p is important for understanding the underlying molecular pathogenesis of ESCC.

The present study demonstrated that both DHRS2 and MYO1B were directly regulated by miR-145-3p in ESCC cells. Overexpression of DHRS2 and MYO1B was observed in ESCC clinical specimens, and overexpression was associated with cancer cell aggressiveness. DHRS2 was initially cloned from a HepG2 human hepatocarcinoma cDNA library and named HEP27 (44). DHRS2 is a member of the short-chain dehydrogenase/reductase (SDR) family that metabolizes many different compounds (45). HEP27 protein interacts with MDM2, which is a negative regulator of p53, resulting in p53 stabilization and induction of p53 transcriptional target genes (46). More recently, downregulation of DHRS2 was reported in ESCC tissues and its downregulation was associated with ESCC aggressiveness and clinical staging (47). Thus, that report arrived at the opposite conclusions from those in our study. Further investigation of DHRS2 function in cancer cells will be necessary.

MYO1B is a member of the membrane-associated class I myosin family and it bridges membrane and actin cytoskeleton in several cellular processes (48). It was recently demonstrated that overexpression of MYO1B is associated with head and neck cancer pathogenesis (20). Importantly, antitumor miR-145-3p directly regulated expression of MYO1B in head and neck cells (20). A previous in vivo study demonstrated that downregulation of MYO1B inhibited cervical lymph node metastasis in head and neck cancer cells (49). These findings indicate that aberrantly expressed MYO1B is associated with cancer cell aggressiveness and metastasis. MYO1B may be a novel diagnostic and therapeutic target for patients with ESCC.

Downstream genes modulated by DHRS2 or MYO1B in ESCC cells were investigated. Previous studies have demonstrated that several aberrantly expressed oncogenes (CDK1, BIRC5, BUB1, TOP2A, CENPF, FOXM1 and AURKA) enhanced cancer cell aggressiveness (38,50-54). Notably, MMP13 may be controlled by DHRS2 and MYO1B in ESCC cells. Our recent study demonstrated that overexpression of MMP13 occurred in ESCC clinical specimens and the expression of MMP13 promoted cancer cell proliferation, migration and invasion (28). Identification of the downstream genes regulated by the miR-145-3p/DHRS2 or miR-145-3p/MYO1B axis may improve the understanding of ESCC aggressiveness.

In conclusion, genes controlled by the antitumor activity of miR-145-3p were closely associated with ESCC pathogenesis. Association of the passenger strand of miRNA is a novel concept of cancer research. DHRS2 and MYO1B were directly regulated by miR-145-3p in ESCC cells. Aberrantly expressed DHRS2 and MYO1B enhanced ESCC cell aggressiveness. Elucidation of antitumor miRNAs controlling RNA networks may provide novel prognostic markers and therapeutic targets for this disease.

Funding

The present study was supported by KAKENHI grants nos. 15K10801, 18K16322, 16K10508, 17K10706, 16K10510, 18K08687, 18K08626 and 17H04285.

Availability of data and materials

All data generated or analyzed during this study are included in this published article.

Authors' contributions

MS and TI performed the majority of the study and wrote the manuscript. NS and SN designed the study and wrote the manuscript. YY, TAra, YK and HK performed the experiments and data interpretation. TAri, KS, IO, YU and KM provided sample collection and clinical support. All authors reviewed, edited and approved the final version of the manuscript.

Ethics approval and consent to participate

The present study was approved by the Bioethics Committee of Kagoshima University (Kagoshima, Japan; approval no. 28-65). Written prior informed consent and approval were obtained from all patients.

Patient consent for publication

All patients had provided written informed consent prior to surgery.

Competing interests

The authors declare that they have no competing interests.

Acknowledgments

Not applicable.

References