Human esophageal squamous carcinoma cell lines (KYSE520 and 790) were established by Shimada and colleagues, and cultured in Ham’s F12/RPMI-1640 with 2% FCS according to a previously reported method (9,10).

cDNAs encoding Hsa-miR-203 sequences were subcloned into a pBApro-CMV Neo DNA plasmid (Takara Bio Inc., Shiga, Japan) in which a neomycin selectable marker was encoded. The hsa-miR-203 plasmid or mock plasmid was transfected into KYSEs with lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to a standard protocol (Lipofectamine 10 μl per vector DNA 4 μg). Cells were then cultured with medium containing 400 μg/ml of G418. Colonies were chosen 7 days after transfection and were then cultured.

Adherent cells were trypsinized, washed once in cold PBS, and 2×10⁵ viable cells were resuspended in 50 μl of PBS with 0.5% bovine serum albumin (staining buffer). Cells were incubated with 1 mg/ml of mouse anti-p75NTR (clone NGFR5) for single staining before fluorescence-activated cell sorting analysis (Becton Dickinson, Inc., San Jose, CA, USA). After washing with staining buffer, cells were resuspended in 50 μl of staining buffer and incubated with 1 mg/ml of the FITC-conjugated goat anti-mouse immunoglobulin G (IgG) antibody. Both primary and secondary staining reactions were carried out for 15 min at room temperature. Non-specific isotype matched antibodies were used as controls. DAPI was added to exclude dead cells. Cells were analyzed using BD LSRII (Becton Dickinson) and results were analyzed with Cell Quest software (Becton Dickinson).

Total RNA was extracted from cells using the TRIzol Reagent (Invitrogen) method. Murine tissue (0.5–2 mg) was homogenized using a bench-top homogenizer in 1 ml TRIzol Reagent (Invitrogen).

cDNA was synthesized using the Superscript First-Strand Synthesis System (Invitrogen). Real-time PCR reactions with the QuantiTect SYBR green PCR kit (Qiagen, Maryland, MA, USA) were run on an ABI Prism^® 7300 (Applied Biosystems, Branchburg, NJ, USA). mRNA quantities were analyzed in duplicate and normalized against GAPDH as an internal control. Results were expressed as relative gene expression using the ΔΔCt method. The following primer sequences were used: Involucrin: forward - tcctcctcca gtcaataccc, reverse - gctgatccctttgtgtt; p63: forward - cagacttg ccaaatcatcc, reverse - cagcattgtcagtttcttagc; Bmi1: forward -ggagaccagcaagtattgtccttttg, reverse - cattgctgctgggcatcgtaag.

cDNA was prepared from miRNA samples using the Taq Man microRNA reverse transcription kit on the ABI Prism 7000 real-time PCR system according to the manufacturer’s instructions (Applied Biosystems). Predesigned Taq Man microRNA assays for hsa-miR-203 (Assay ID 000507) was purchased from Applied Biosystems. qRT-PCR was performed using a Taq Man universal PCR master mix, according to the manufacturer’s protocol (Applied Biosystems).

All mouse studies were carried out under the approval of the Institutional Review Board of the University of Toyama (no. S-2008MED-56). KYSE cells (1×10⁶) in 100 μl PBS were injected subcutaneously into the right flank of four week-old female CAnN.Cg-Foxn1nu/CrlCrlj (BALB/c) nude mice (Charles River Laboratories, Yokohama, Japan). Animals were sacrificed at week 8 and tumor tissues were harvested. Parts of the tumor tissues were used for RNA and the rest were fixed in 10% formalin and embedded in paraffin.

Sections (4 μm) were cut from paraffin blocks, deparaffinized with xylene, and rehydrated through a graded alcohol series. These sections were then autoclaved at 121°C in Target Retrieval Solution (Dako Cytomation, Kyoto, Japan) for 5 min and cooled to room temperature to unmask antigens. Immunostaining was performed with Envision Plus kits/horseradish peroxidase/3,3’-diaminobenzidine (Dako Cytomation) according to the manufacturer’s instructions. Primary antibodies were incubated at room temperature for 1 h, or at 4°C overnight. Slides were counterstained with Mayer’s hematoxylin. The primary antibodies used in this study were as follows: rabbit polyclonal anti-Ki67, clone ab66155 (Abcam Inc., Cambridge, MA, USA) dilution1:200, mouse monoclonal anti-involucrin, clone SY5 (Santa Cruz Biotechnology, Inc., CA, USA) dilution 1:50, mouse monoclonal anti-p63, clone 4A4 (Abcam Inc.) dilution 1:1, mouse monoclonal anti-Bmi1, clone 1.T.21 (Abcam Inc.) dilution 1:100, mouse monoclonal anti-p75NTR, clone NGFR5 (Santa Cruz Biotechnology, Inc.) dilution 1:50, and rabbit polyclonal anti-Laminin (Abcam Inc.) dilution 1:100.

A 3D-Gene Human Oligo chip 25 k (Toray Industries, Tokyo, Japan) was used (25,370 distinct genes) according to the supplier’s protocols (www.3d-gene.com).

Written informed consent for this research was obtained from patients prior to surgery with approval by the Institutional Review Board (no. 20–75). From a total of 100 serial ESCC patients who had undergone esophagectomy between 1997 and 2007 in the University of Toyama Hospital, RNA was obtained with appropriate quality from the FFPE samples of 31 patients. Then patients with disease stage I or IV and/or patients who received non-radical surgery and/or patients who died within 6 months of surgery were excluded from this study. Finally, in total 20 cases were analyzed in this study. Information on gender, age, stage of disease and histopathological factors was abstracted from the medical records. All of the tumors were confirmed as ESCC by the Clinicopathological Department of the hospital. All cases were classified according to the International Union Against Cancer TNM Classification 7th edition (11).

Sections (10 μm) were prepared from each FFPE specimen. Paraffin was removed by treatment with xylene and tissues were washed with ethanol twice to remove xylene. Tissues were then treated with proteinase K at 37°C overnight. Following centrifugation, the supernatant was processed with a silica-based spin column (Toray Industries) in order to obtain purified total RNA. The degrees of RNA cross-linking and RNA degradation were analyzed by electrophoresis using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

miRNA profiling was examined using a Toray 3D-Gene^® miRNA oligo chip (Toray Industries) on which 885 genes were mounted. The detailed procedure of the experiment has been previously described (12). We defined tumors with high miR-203 expression when the relative expression level of miR-203 was 2-fold higher than that in normal counterparts from the same resected specimen.

mRNA expression levels in mouse tumor and cultured cells were analyzed by the t-test. The Kaplan-Meier method was used to estimate patient survival. Differences in survival for the overexpression (T/N ratio >2.0) of miR-203 were analyzed using the log-rank test. P<0.05 was set as significance.

To investigate the functional role of miR-203 in ESCC, we established stable clones constitutively expressing miR-203 by transfecting the miR-203 plasmid into two ESCC cell lines (KYSE520 and KYSE790). We established a mock transfectant and 2 stable transfectants each from KYSE520 and KYSE790, namely, KYSE520 mock, KYSE520 miR203-1, KYSE520 miR203-2, KYSE790 mock, KYSE790 miR203-1 and KYSE790 miR203-2. miR-203 expression levels in miR203 transfectants were 7–14-fold higher than those in the parent cell and mock transfectant established from KYSE520 (Fig. 1A), whereas, miR-203 expression levels in KYSE790 miR203 transfectants were 1.5–2.1-fold higher than those in the parent cell and mock transfectant (Fig. 1A). Morphologically, the parent cell and the mock transfectant had a homogeneous small angular shape, while miR-203 transfectants were heterogeneous in size and shape, with large round cells with a decreased nuclear-cytoplasmic ratio (Fig. 1B). Cell proliferation was significantly lower in miR-203 transfectants than in the parent cell and the mock transfectant (Fig. 1C). Involucrin expression levels were 8–27-fold higher in the miR-203 transfectants than in the parent cell and mock transfectant (Fig. 1D). Taken together, the induced expression of miR-203 resulted in cell growth inhibition and terminal differentiation in ESCC cell lines.

To investigate the miR-203 induced phenotypic change in specific cell subsets in ESCC, we focused on the expression of p75NTR, which has been reported to be expressed exclusively in basal cells of normal esophageal epithelia (13). There was no significant difference among KYSE520, KYSE520-Mock and the miR-203 transfectants in the proportion of p75NTR-bright cells detected by flow cytometry (Fig. 2A). We isolated the p75NTR bright/dim fraction from the parent, mock and miR-203 stable clones, respectively (Fig. 2B). In the parent and mock transfectant, the relative expression level of miR-203 in p75NTR dim cells was higher than that in p75NTR bright cells (Fig. 2C). miR-203 expression was induced in both the p75NTR bright and dim cells of the miR-203 stable clone (Fig. 2C), which indicated that miR-203 expression was broadly induced in the stable clone by the vector with the CMV promoter. Then we detected the expression of p63 and Bmi-1, which have been reported to regulate the self-renewal and differentiation of normal epithelial stem cells (14–16), by RT-PCR. In the miR-203 stable clone, p63 expression was higher in the p75NTR bright fraction than in the p75NTR dim fraction, while Bmi1 was expressed to the same extent in both the p75NTR bright and dim fractions (Fig. 2D). These results indicated that suppression of p63 in the p75NTR dim fraction was involved in the miR-203 induced phenotypic change in vitro.

We subcutaneously injected miR-203 stable clones into mice (n=3 in each clone). Four weeks after the injection, tumors were established from all the cell lines including miR-203 stable clones (Fig. 3A), which indicated that the induction of miR-203 did not affect the tumor-forming efficacy of the cells. On the other hand, the size of the tumors established from injecting miR-203 stable clones was significantly smaller than that from the parent and mock transfectant (Fig. 3B). Histological examination of H&E-stained sections of the tumors showed small cancer cells homogeneously distributed in the tumors established from injecting the parent and mock transfectant (Fig. 3D). In contrast, multi-centric foci were seen in the tumors established from injecting miR-203 stable clones (Fig. 3D). The cells were large, eosinophilic, and polygonal, and appeared to be stratified from the center of the foci toward the outer layer in which cornification was seen, resembling the architectural pattern of squamous epithelia.

Ki67 staining revealed that proliferating cells were broadly distributed in tumors established from the parent and mock transfectant, while proliferating cells were limited to the center area of the foci in tumors from miR-203 stable clones (Fig. 3E). The ki67 labeling index was significantly lower in tumors from miR-203 stable clones than in the parent and mock transfectant (Fig. 3C). Very weak involucrin expression was observed in tumors from the parent and mock transfectant, while involucrin was expressed in cornified cells at the outer layer of the foci in tumors from miR-203 stable clones (Fig. 3F).

Detailed high power observations of serial sections of tumors from miR-203 stable clones showed that small spindle-shaped cells lined the center of the foci as if to mimic cells on the basal layer in squamous epithelia. Round cells lined the middle layers, and flat cornified cells lined the outer layers (Fig. 3G). Ki67 was expressed on cells at the innermost layers (Fig. 3G), and involucrin was expressed in cornified cells at the outer layers (Fig. 3G). Laminine, a basal membrane protein, was expressed in the inner area of the innermost layer (Fig. 3H). p75NTR expression was limited to a small number of cells scattered on the innermost layer (Fig. 3I). The immunohistochemical detection on serial sections indicated recovered epithelial tissue polarity with basement membrane formation upon the induction of miR-203 in vivo.

To investigate the molecular mechanism involved in the miR-203-induced phenotypic change, we profiled gene expression in KYSE520 mock and KYSE520 miR203-2 using cDNA microarray analysis (TORAY-3D gene chip with 25,370 distinct genes). The upregulation of a total of 1539 genes was >2-fold higher in the miR-203 transfectant than in the mock transfectant (data not shown). Genes encoding basal membrane proteins such as laminin and collagen type 4 were markedly upregulated in the miR-203 transfectant (Table I). Genes involved in regulating cell polarity and tissue architecture, such as Rock-1, BMP-4 and ZO-1, were also upregulated in the miR-203 transfectant (Table I).

Since p63 and Bmi-1 have been reported to be the major targets of miR-203 and to regulate stem cell properties (7,8,17–19), we detected the protein expression of these molecules by immunohistochemistry. p63 expression was broadly distributed in tumors established from the parent and mock transfectant, while expression was limited in the first several layers at the center area of the foci in tumors from miR-203 stable clones (Fig. 4A). Bmi-1 was expressed in almost all cells and was broadly distributed in tumors even those from miR-203 stable clones (Fig. 4B). These results indicated that miR203-induced suppression of p63 in supra-basal cell fraction was involved in the phenotypic change in this experiment, while Bmi-1 did not play a key role in the previous model.

To assess the clinical importance of miR-203 on ESCC, we next investigated the relationship between miR-203 expression in surgically removed ESCC specimens and clinicopathological characteristics of the patients.

In the 20 specimens analyzed in this study, miR-203 expression was high in 7 specimens (35.0%) (the relative expression level of miR-203 in tumors was >2-fold higher than that of normal counterparts) and low in 13 specimens (75.0%). No significant correlation was observed between miR-203 expression and various factors such as age, gender, tumor location, extent of the tumors, lymph node metastasis, distant metastasis, pTNM pathological classification or histology (Table II). However, Kaplan-Meier survival curves revealed that miR-203 expression significantly correlated with a favorable outcome (Fig. 5). Multivariate analysis revealed that the expression of miR-203 was an independent prognostic factor (P=0.0006).