This study protocol was approved by the Medical Ethics Committees of Xi'an Jiaotong University of Medicine (no. H34-32-1, Xi'an, China). All investigations were conducted in accordance with the Declaration of Helsinki. All the patients involving this study provided a written informed consent. The nude mice used in this study were treated in accordance with the institutional guidelines of the Animal Ethics Committee for the care and use of animals in Xi'an Jiaotong University of Medicine, China. We confirm that all methods were carried out in accordance with relevant guidelines and regulations of Xi'an Jiaotong University of Medicine.

We previously described the method of CT culture (16). Briefly, 21 samples of cervical cancer tissue (stage IB, n=14; stage IC, n=4; stage IIa, n=3; age, 36–69 years) were obtained by resection (Table I). For CT culture, tumors were washed immediately in phosphate-buffered saline (PBS) containing 500 U/l penicillin G (Gibco; Thermo Fisher Scientific, Waltham, MA, USA) and 500 mg/l streptomycin (Gibco) <30 min after resection, and then digested overnight in DMEM/F12 medium supplemented with 0.5 mg/ml collagenase IV (Gibco). Tumors were cultured in stem cell medium (DMEM/F12), 10 ng/ml basic fibroblast growth factor (bFGF), 10 U/ml leukemia inhibitory factor, 1×10⁵ U/l penicillin, 100 mg/l streptomycin (all reagents from EMD Millipore, Merck KGaA, Darmstadt, Germany) at 37°C in a humidified atmosphere containing 5% CO₂. Clones of >50 cells were recognized as tumorspheres. These were dissociated every 7–10 days by incubation in a non-enzymatic cell dissociation solution (Sigma-Aldrich, St. Louis, MO, USA) for 2 min at 37°C and passaged at 1×10³ cells per 100-mm plate. Tumorsphere cells had differentiated completely by 8 days after switching to stem cell medium without bFGF.

All Ad-vectors had comparable titers of 10⁸–10⁹ transducing units/ml. Virus suspensions were stored at −80°C until use. Suspensions were centrifuged briefly and kept on ice immediately before use. For transduction, 2×10⁴ dissociated tumorsphere cells were transduced 1 day after initial seeding of cells with a multiplicity of infection (MOI) of 25. Cells were incubated in stem cell medium containing adenovirus particles and 4 µg/ml polybrene (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) for 18 h at 37°C in a humidified atmosphere containing 5% CO₂. Ad-particles were removed and the medium replaced with fresh stem cell medium.

A colony formation assay was carried out as previously described (19). Briefly, single cervical carcinoma cells or dissociated tumorsphere cells were cultured in DMEM/F-12 medium supplemented with 10% fetal calf serum (FCS; Life Technologies, Carlsbad, CA, USA), 2 mmol/l glutamine (Invitrogen, Carlsbad, CA, USA), 1×10⁵ U/l penicillin, and 100 mg/l streptomycin. Cells were cultured at clonal densities of 100–300/cm² on 2% gelatin (Sigma-Aldrich)-coated tissue culture dishes (BD Biosciences, San Jose, CA, USA) at 37°C in 5% CO₂ in air. Cloning plates were monitored every day and medium was changed every 2–3 days. After 28 days of culture, plates were fixed in 10% formaldehyde/PBS for 10 min and stained with Harris hematoxylin. Clones (>50 cells) were counted on ≥3 plates per sample and averaged. Efficiency of colony formation was determined as [(number of colonies)/(number of cells seeded)] ×100.

Dissociated tumorsphere cells or adenovirus vector-transfected cells were grown in 6-well plates and collected. Cells were resuspended in PBS containing 2% fetal bovine serum (FBS) and 0.1% sodium azide, and then analyzed by a FACScalibur system (BD Biosciences). Acquisition was set for 10,000 events per sample. Data were analyzed using FACS v4.1.2 (BD Biosciences). Triplicate samples were analyzed in each experiment.

Total RNA was extracted from cells using TRIzol® (Invitrogen). miRNA levels were assayed using Taqman® probes and primer sets (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions. Briefly, first-strand cDNA was generated using a reverse transcription system kit (Promega, Madison, WI, USA) with random primers of miR-145. Realtime PCR was carried out using power SYBR Green PCR master mix (Applied Biosystems) in a StepOnePlus® system (Applied Biosystems). The level of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA was used as the control for internal normalization. For exact quantification of gene copies per cell, reverse-transcribed miR-145 cDNA was used as a template to formulate standard curves. Then, the exact number of copies of miR-145 per cell was calculated according to their molecular weight and cell counts. Primer sequences are presented in Table II.

After dissociation of tumorspheres and tumorsphere-derived differentiated cells from 21 cervical cancer patients in a non-enzymatic cell-dissociation solution, cells were washed in serum-free Hank's balanced salt solution (HBSS). Then, cells were suspended in a 1:1 (v/v) mixture of serum-free DMEM/F12, and 1×10⁵ cells were injected (s.c.) into the right (differentiated cells) or left (tumorsphere cells) mid-abdominal area of nude mice using a 23-G needle. Animals were subjected to necropsy 28 days after implantation, and tumor growth assessed by measuring volume using the formula: V = 1/2 × (L × W²). In the silenced-miR-145 group, mice were sacrificed at 14 days after injection to avoid necrosis in transplanted tumors. Sequences of primary miR-145 or silent miR-145 are shown in Table III.

Lysates were extracted from tumorspheres or cells. Proteins (20 µg) were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto nitrocellulose membranes. The latter were blocked in 5% skimmed milk in Tris-buffered saline-Tween-20, and then incubated with primary antibodies to core TF proteins or GAPDH (1:500 dilution; Santa Cruz Biotechnology, Inc.) overnight at 4°C. After incubation with horseradish peroxidase-labeled secondary antibodies, membranes were developed using a SuperSignal® West Pico Trial kit (Pierce, Rockford, IL, USA).

Cervical cancer samples were fixed in phosphate-buffered 10% formalin (pH 7.2), embedded in paraffin, and cut into sections (4 µm). Sections were dewaxed in xylene, dehydrated in alcohol, and incubated in 0.01 M sodium citrate buffer (pH 6.0) for antigen retrieval. Sections were incubated with 3% H₂O₂ for 30 min to block endogenous peroxidase activity and with normal mouse serum at 37°C for 15 min to block non-specific binding of antibody. Then, sections were incubated with Sox2, Oct4 or Nanog antibodies (1:100 dilution in PBS; Santa Cruz Biotechnology, Inc.) for 2 h at room temperature, followed by incubation with biotinylated secondary antibody (Santa Cruz Biotechnology, Inc.) and 3,3′-diaminobenzidine.

The complementary DNAs of Sox2, Nanog and Oct4 were purchased from Shanghai Yingji Biological Company (Shanghai, China). Mutated constructs containing the 6 bp point mutations in the core TF seed sequence predicted to be the miR-145 target sequence were synthesized using a QuikChange II site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA). The primers were designed by Stratagene using their own software (http://labtools.stratagene.com/QC) and are shown in Table IV. Ad-vectors expressing the core TF wild-type or mutation of miR-145 seed sequence under the control of the U6 promoter were produced by transient transfection of HEK293T cells, and then transfected into tumorspheres following the procedures described above.

Titers of virus stocks were checked using TCID₅₀ methods. High-titer stocks were stored at −70°C until use. Two weeks after injection of tumorspheres into nude mice, the tumors became palpable and each tumor site was injected with 5.8×10⁵ pfu of Ad-miR-145 and Ad-Mock with gadolinium (1:1) (Sigma-Aldrich). Tumor size was measured each week.

Cell invasion was evaluated using 24-well Transwell® culture chambers, as previously described (18). Cells were seeded at 5×10⁴ per well and cultured in stem cell medium for 24 h. Then, cells were fixed in methanol and stained with 5% crystal violet. After examination under a light microscope, cells were eluted with 33% acetic acid. Optical-density values of the eluate were read using a Bio-Rad microplate reader at 590 nm.

The protocol followed that previously described (20). Briefly, data on expression of genes and miRNA were obtained from the Cancer Genome Atlas (TCGA; https://tcga-data.nci.nih.gov/tcga/tcgaHome2.jsp) for patients diagnosed with squamous cervical cancer. Expression of genes or miR-145 was divided into 'high' and 'low' groups based on the mean ± 1 SD. Kaplan-Meier survival curves were generated comparing these two groups via the log-rank test. miR-145 expression combined with expression data for Oct4, Nanog, or Sox2 were subtracted from the miR-145 data for each patient, and the analysis described above was repeated.

Data are shown as the mean ± standard error of the mean. Comparison between groups were performed using analysis of variance, Fisher's exact test or two-tailed Student's t-test. P<0.05 was considered to indicate a statistically significant difference.

Our results showed that cervical cancer tissues were immunopositive for Nanog (20/21), Oct4 (19/21) and Sox2 (20/21), and 19/21 samples were positive for all three TFs. Staining for these TFs was visible in the cytoplasm and nuclei of a few (but not all) cancer cells, and we could detect that most cervical cancer cells expressed these TFs (18/21) (Fig. 1A). After enzymatic dissociation and culture in stem cell medium, a few colonies (tumorspheres) formed in all tissues tested (21/21) after 15 days. Colonies were formed at the rate of 22.41±3.23 per 100 cells dissociated from tumorspheres, but this rate was reduced to only 0.41±0.07 per 100 cells dissociated from differentiated tumorspheres (P<0.001) (Fig. 1B) when they detached from the disks. Furthermore, protein levels of Nanog, Sox2 and Oct4 were significantly higher in tumorsphere cells than in differentiated cells (P=0.012, Fig. 1C).

Next, we examined expression of miR-145 and core TFs according to tumorsphere stages. First, we assessed expression of miR-145 in cells cultured under tumorsphere conditions or differentiated conditions [removal of basic fibroblast growth factor (bFGF) from the stem cell medium]. Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) showed that the protein levels of core TFs decreased, whereas miR-145 level increased, at different time points and underwent dynamic changes (Fig. 1D and E). Two studies previously reported potential miR-145 binding sites in the genes of these TFs (19,21), so we used miRanda software to predict miR-145 binding sites in the mRNAs of these TFs, the results confirmed potential miR-145 binding sequences within them (Fig. 1F).

Next, we investigated the role of miR-145 in cervical cancer. Ad-miR-145-GFP was transduced into tumorspheres or their differentiated cells to increase miR-145 expression. Flow cytometric analyses showed that there were 91±1.9% GFP-positive cells in the overexpressed Ad-miR-145-GFP group and 91±2.3% in the adenovirus-scrambled sequence-GFP (Ad-scr) group with no homology to the human genome, which acted as the control group among transfected tumorsphere cells. The miR-145 level was increased 4.5-fold in the overexpressed Ad-miR-145-GFP group compared with its control (Fig. 2A). Accordingly, levels of core TFs were decreased significantly according to western blotting, and colony formation and cell invasion was decreased in the overexpressing Ad-miR-145-GFP group (Fig. 2B–D). Next, dissociated tumorsphere cells harboring Ad-miR- 145-GFP or Ad-scr-GFP were injected into nude mice. A total of 21 Ad-scr-GFP-tumorspheres derived from the 21 patients caused tumor formation in mice after 28 days. In contrast, 21 Ad-miR-145-GFP-tumorspheres from these 21 patients only caused 11 tumors, and those tumors were relatively small compared with those in the Ad-scr-GFP group (Fig. 2E).

Next, we transfected the Ad-silence-miR-145 (Ad-si-miR-145-GFP) vector or its control (Ad-scr-GFP) into CTs. Flow cytometric analyses showed that there were 90±1.4% GFP-positive cells in the Ad-si-miR-145-GFP group and 88±3.2% in the Ad-scr-GFP group. The miR-145 level was decreased fourfold compared with the control (Fig. 3A), whereas levels of TFs, colony formation, and cell invasion increased after silencing of miR-145 (Fig. 3B–D). Tumor formation was 21/21 in the Ad-si-miR-145 group and 20/21 in the scr-GFP group. To prevent tumor growth leading to necrosis (which would have affected the measurement of tumor volume), mice were sacrificed 14 days after injection. Tumors were larger in the si-miR-145 group than in the scr-GFP group (Fig. 3E).

miR-145 seed sequences were predicted within Oct4, Sox2 3′-untranslated regions (UTR) and Nanog coding sequence (CDS) by the software miRanda after mutation of miR-145 target sequences (6 bp deletion of the miR-145 target site) in these core transcription factors (Fig. 4A), and the vectors for the mutated structures are shown in Fig. 4B. First, we transfected Ad-Sox2-EGFP, Ad-Sox2-EGFP-mutation (mut)-miR-145 target sequence (Ad-Sox2-EGFP-mut) into tumorspheres, a luciferase reporter with no UTR of Sox2 and Oct4, CDS of Nanog were used as a negative control. Flow cytometric analyses revealed the percentage of EGFP-positive cells to be 86-89% (87±2.2%), 88-89% (88±0.6%), and 87-89% (88±1.9%), respectively. The Sox2 level did not significantly differ between the wild-type (WT) group and the control (con) group by 24 or 48 h after transfection (P=0.47 and 0.64, respectively), but Sox2 expression in the WT group was lower than in the other two groups (P=0.012 and 0.011, respectively) (Fig. 4C). Similarly, following transfection of Ad-Oct4-EGFP, Ad-Oct4-EGFP-mut and Ad-control-EGFP, or Ad-Nanog-EGFP, Ad-Nanog-EGFP-mut and Ad-control-EGFP, flow cytometric analyses revealed the percentage of GFP-positive cells to be 83–86% (84±1.9%) and 83–87% (85±1.7%), respectively. The Oct4 and Nanog levels did not differ between the mut group and the control group (P=0.75 and 0.57, respectively), and their expression was lower in the WT groups than in the other groups (P=0.002 and 0.005, respectively) (Fig. 4D and E).

We further addressed the dependence of the reported repression of Oct4, Sox2 UTR reporter, and Nanog CDS reporter on the level of miR-145. After depletion of miR-145 by antisense inhibitor, locked nuclei acid (LNA) in CT completely abolished the differential regulation between the mutant and WT 3′-UTR reporters in Oct4 and Sox2 (Fig. 4F–H), and the mutant and wild-type CDS in Nanog. These results were in accordance with the reports by Xu et al (13) and Wang et al (21).

Tumor cells were dissociated from tumorspheres and injected into null mice. The resulting tumors were visible or palpable 2 weeks after injection. Next, we injected 5.8×10⁵ pfu of Ad-miR145 or Ad-Mock with gadolinium combination into tumors three times a week (22). Mice were sacrificed at day 28 and tumor volume measured. Tumors were smaller in the Ad-miR-145 group than in the AD-scr group (P=0.009, Fig. 5A), the weak expression of core stem cell TFs was detected after Ad-pri-miR-145 treatment (Fig. 5B). Moreover, levels of core TFs were decreased (P=0.03, Fig. 5C), miR-145 level was increased (P=0.04, Fig. 5D), colony formation was decreased (P=0.007, Fig. 5E), and cell invasion was decreased (P=0.012, Fig. 5F) in the Ad-miR-145 group compared with the control group.

To ascertain the clinical relevance of miR-145 and these core stem cell transcripts in the prognosis of CC patients, we evaluated the TCGA dataset. We found that high expression of miR-145 indicated a better prognosis compared with low expression of miR-145 (Fig. 6A). When miR-145 levels were combined with Nanog expression from the same patient, patients in the low Nanog/high miR-145 expression group had a better prognosis compared with the high Nanog/low miR-145 expression group (Fig. 6B). Similar results were obtained for expression of low miR-145 with high Sox2 and Oct4 in these subtypes compared with their opposite groups. These results implied that these genes could be informative for survival of CC patients (Fig. 6C and D).