The experiment was performed according to previously described methods (21). Briefly, surgically isolated tissues from 4 ovarian cancer patients (Table I) were minced, digested with 0.25% trypsin (Gibco; Thermo Fisher Scientific, Inc.), and centrifuged at 4°C, 450 × g for 5 min. The cell precipitates were collected and incubated with mouse anti-human CD133-FITC antibodies (1:100; cat. no. 11-1339-42; eBioscience; Thermo Fisher Scientific, Inc.) and rabbit anti-human CD44-APC antibodies (1:100; cat. no. 17-0441-82; eBioscience; Thermo Fisher Scientific, Inc.) at 4°C for 30 min. Next, CD44⁺/CD133⁺ OCSCs were sorted from the sample by fluorescence-activated cell sorting. After sorting, single cells were plated at 1,000 cells/ml in DMEM/F12 (Gibco; Thermo Fisher Scientific, Inc.), supplemented with 10 ng/ml basic fibroblast growth factor, 10 ng/ml epidermal growth factor, 5 µg/ml insulin and 0.5% BSA (all from Sigma-Aldrich; Merck KGaA, and cultured at 37°C with 5% CO₂. Cells of the fourth passage were used to perform subsequent experiments. The study involving human tissues was approved by the Ethics Review Committee of Shanghai Geriatric Institute of Chinese Medicine of Research in Human Production, authorized by Shanghai Municipal Government; written informed consent was provided by all patients, in accordance with The Declaration of Helsinki.

Anisomycin (purity >98%) and dimethyl sulfoxide (DMSO) were purchased from Sigma-Aldrich (Merck KGaA). According to the IC₅₀ value determined by MTT assay, anisomycin was applied at a final concentration of 31.8 mM in all the assays. As a control, the same volume of DMSO was applied to the cells.

Briefly, all fresh tissues were soaked in 4% paraformaldehyde (Sigma-Aldrich; Merck KGaA) at room temperature for 30 min. The tissues were dehydrated using an ethanol gradient and embedded in paraffin. Then, the tissues were sectioned (thickness, 6 µm) and soaked in xylene for dewaxing. The tissue sections were stained with haematoxylin and eosin (Sigma-Aldrich; Merck KGaA) and were finally cleared with xylene and mounted with neutral resin (Sigma-Aldrich; Merck KGaA). Five random fields of each tissue section from a total 4 slides were observed (magnification, ×200) with a light microscope (Olympus CX23; Olympus Corporation).

Briefly, all fresh tissues were soaked in 4% paraformaldehyde (Sigma-Aldrich; Merck KGaA) at room temperature for 30 min. The tissues were dehydrated using an ethanol gradient and embedded in paraffin. Then, the tissues were sectioned (thickness, 6 µm) and soaked in xylene for dewaxing. Next, sections were rinsed with 3% phosphate buffer (Sigma-Aldrich; Merck KGaA). Mouse anti-human CD34 and CD31 primary antibodies (1:200; cat nos. 3569 and 3528, respectively; Cell Signaling Technology, Inc.) were added for 60 min at room temperature, and then anti-mouse horseradish peroxidase-conjugated secondary antibody (1:200; cat no. 7076; Cell Signaling Technology, Inc.) was added for 60 min at room temperature. Finally, ABC chromo-genic reagent (VECTASTAIN Elite ABC kit; cat no. PK-6100; Vector Laboratories, Inc.) was used for the color reaction. PBS (pH 7.4) was used as a negative control in the place of the primary antibody. Five random fields of each tissue section were observed (magnification, ×200) and analyzed using IPP software (Image-Pro Plus Version 6.0; Media Cybernetics Co. Ltd.). ImageJ 1.42q (National Institutes of Health) was used to analyze the area of the positive cells on IHC staining.

miR-421 and miR-421 mutant (miR-mut; a miRNA mimics negative control) oligonucleotide RNAs were purchased from Shanghai GenePharma Co., Ltd. HuOCSCs were transfected with 0.2 mg miR-421 mimics or miR-mut using Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's protocol. The cells were used for subsequent experiments at 24 h post-transfection.

Briefly, a total 1,000 cells from each group in 100 µl medium were seeded in a 96-well plate. After 24 h, 10 µl of MTT solution (Sigma-Aldrich; Merck KGaA) was added to each group of cells for incubation at 37°C for 3 h. The medium was discarded, 150 µl of DMSO was added to each well, and the plate was shaken for 15 sec to mix well. The culture plate was placed in a microplate reader to record the absorbance value at 450 nm. The formula for calculating the cell proliferation inhibition rate (%) was: [1-(OD value of experimental group of cells-blank)/(OD value of control group of cells-blank)] ×100.

The experiment was performed according to the instruction manual of the Annexin V-FITC Apoptosis Detection kit (Beyotime Institute of Biotechnology). Briefly, adherent cells were digested using trypsin. The cells were washed with PBS once, centrifuged, and gently resuspended in 195 µl of Annexin V-FITC binding solution. Next, 5 µl of Annexin V-FITC was added, and the sample was gently mixed. Finally, 10 µl of PI staining solution was added, and the sample was gently mixed and incubated at 20°C in the dark for 30 min. The cells were then detected using a flow cytometer (Cytomics FC 500; Beckman Coulter, Inc.) and analysed using FlowJo software (Version 7.6.1; FlowJo, LLC).

The upper chambers of the tran-swell filters (8.0-µm pore, 6.5-mm polycarbonate; Corning, Inc.) were pre-coated with 50 µl Matrigel (Corning, Inc.). Cells (2×10⁵) were resuspended in 200 µl serum-free DMEM/F12 and seeded on the upper chambers of the Transwell filters. A total of 600 µl DMEM/F12 containing 10% FBS was added to the bottom chambers. Cells were allowed to invade for 24 h at 37°C in a humidified incubator with 5% CO₂. Cells invaded to the lower surface of the filter were fixed in 4% paraformaldehyde at room temperature for 30 min, and stained with DAPI (Beyotime Institute of Biotechnology). Invaded cells were then counted using a fluorescent microscope (Olympus CX23; Olympus Corporation). Five random optical fields were counted for each Transwell filter.

All steps were performed according to a previously described method (22). In brief, human umbilical vein epithelial cells (HUVECs) were divided into three experimental groups: the non-treated group, the anisomycin-treated group (anisomycin was applied at a final concentration of 31.8 mM; dilution for 1:1,000) and the vehicle-treated group (the same volume of DMSO was applied to the cells). Each group cells was plated on Matrigel-coated 6-chamber slides (2×10⁵ cells/chamber), in the presence or absence of treatments as aforementioned. After 6 h of incubation, the cells were photographed using a light microscope (Olympus CX23; Olympus Corporation). To quantitate the data, the number of branch points in four non-overlapping fields was counted.

Female BALB/c nude mice (n=4; weight, ~20 g; age, 8 weeks) were obtained from the Shanghai Research Center for Model Organisms (Shanghai, China). This study was approved (permit no. SRCMR20160018) by the Animal Ethics Committee of Shanghai Research Center for Model Organisms, and the experimental protocols were in compliance with the Experimental Animal Regulations of the Ministry of Science and Technology National Science and Technology Commission (Beijing, China). All mice were maintained in colonies of 1 per cage for 7 days, and housed in a temperature-controlled room under a standard light-dark cycle, with free access to food and water. Two types of cells HuOCSCs were inoculated into the back of all mice: DMSO-pretreated cells (total 1×10⁵ cells in 100 µl per mouse) were inoculated into the scapula and anisomycin-pretreated cells (total 1×10⁵ cells in 100 µl per mouse) were inoculated into the lower back. Each experimental group comprised four mice. After 35 weeks, the mice were sacrificed and the tumors were surgically excised. Tumors were weighed, and the volume was calculated as follows: Tumor volume (mm³) = [longest axis (mm) x shortest axis (mm)²]/2.

For the histological analysis of the xenograft tumors, the fresh tumor tissues were soaked in 4% paraformaldehyde (Sigma-Aldrich; Merck KGaA) at room temperature for 30 min. The tumor tissues were dehydrated using an ethanol gradient and embedded in paraffin. Then, the paraffin blocks were sectioned (thickness, 6 µm) and soaked in xylene for dewaxing. The tumor tissue sections were stained with hematoxylin and eosin (Sigma-Aldrich; Merck KGaA) and were finally cleared with xylene and mounted with neutral resin (Sigma-Aldrich; Merck KGaA). Three random fields of each tumor tissue section from a total 4 slides were observed (magnification, ×200) using a light microscope (Olympus CX23; Olympus Corporation). In addition, immunofluorescence analysis was performed. The tumor tissue sections were rinsed with 3% phosphate buffer, and mouse anti-CD31 (clone P2B1; cat. no. ab24590; Abcam), rabbit anti-CD146 (clone EPR3208; cat. no. ab75769; Abcam), mouse anti-coagulation factor VIII (FVIII; clone GMA-012; ab78852; Abcam) and rabbit anti-fms related tyrosine kinase 1 (Flt1; clone Y103; cat. no. ab32152; Abcam) were added at 1:200 dilution for 60 min at room temperature. Then, the tissue sections were incubated with goat anti-mouse (Alexa Fluor^® 555 pre-adsorbed; 1:200; ab150118; Abcam) or goat anti-rabbit secondary antibody (Alexa Fluor^® 488 pre-adsorbed; 1:200; ab150077; Abcam) and 5 µg/ml DAPI at room temperature for 60 min. The tumor tissue sections were then washed thoroughly with TBST [25 mM Tris/HCl (pH 8.0), 125 mM NaCl and 0.05% Tween-20] and observed under a fluorescence microscope (DMI3000; Leica Microsystems, Inc.). PBS (pH 7.4) was used as a negative control in the place of the primary antibodies. Three random fields of each tissue section were observed (magnification, ×200) and analyzed using IPP software (Image-Pro Plus Version 6.0; Media Cybernetics Co. Ltd.). ImageJ 1.42q (National Institutes of Health) was used to analyze the area of the positive cells on the immunofluorescence staining.

Total RNA from each group of cells was extracted using TRIzol reagent (Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions. Total RNA was treated with DNase I (Sigma-Aldrich; Merck KGaA), quantified and reverse transcribed into cDNA using the ReverTra Ace-α First Strand cDNA Synthesis kit (Toyobo Life Science). qPCR was performed with a RealPlex4 real-time PCR detection system from Eppendorf Co. Ltd. SYBR Green Real-Time PCR Master Mix (Toyobo Life Science) was used as the fluorescent dye. qPCR was performed as follows: 95°C for 5 min, then 40 cycles of denaturation at 95°C for 15 sec, annealing at 58°C for 30 sec and extension at 72°C for 42 sec, followed by a final extension step at 72°C for 10 min. The relative gene expression levels were calculated using the 2^−ΔΔCq method (ΔCq=Cq of gene -Cq of 18s rRNA; ΔΔCq=ΔCq of experimental group-ΔCq of control group) (23). The mRNA expression levels were normalised to the expression levels of 18s rRNA. The primers for amplification of each gene were as follows: Notch1, forward 5′-CGC TGA CGG AGT ACA AGTG-3′ and reverse 5′-GTA GGA GCC GAC CTC GTT G-3′; d-like canonical Notch ligand 1 (DLL1), forward 5′-TGT GAC GAG TGT ATC CGC TAT-3′ and reverse 5′-GTG TGC AGT AGT TCA GGT CCT-3′; hes family bHLH transcription factor 1 (HES1), forward 5′-CCT GTC ATC CCC GTC TAC AC-3′ and reverse 5′-CAC ATG GAG TCC GCC GTA A-3′; recombination signal binding protein for immunoglobulin κ J region (RBPJ), forward 5′-AAC AAA TGG AAC GCG ATG GTT-3′ and reverse 5′-GGC TGT GCA ATA GTT CTT TCC TT-3′; PDGFRA, forward 5′-TTG AAG GCA GGC ACA TTT ACA-3′ and reverse 5′-GCG ACA AGG TAT AAT GGC AGA AT-3′; platelet derived growth factor receptor β (PDGFRB), forward 5′-TGA TGC CGA GGA ACT ATT CAT CT-3′ and reverse 5′-TTT CTT CTC GTG CAG TGT CAC-3′; miR-421, forward 5′-ATC AAC AGA CAT TAA TTG GGC GC-3′ and reverse 5′-GCT GTC AAC GAT ACG CTAC CTA-3′; miR-145, forward 5′-GTC CAG TTT TCC CAG GAA TCC CT-3′ and reverse 5′-GCT GTC AAC GAT ACG CTA CCT A-3′; lncRNA-Meg3, forward 5′-GCC CAT CTA CAC CTC AC-3′ and reverse 5′-ATC CTT TGC CAT CCT G-3′; and 18S rRNA, forward 5′-CAG CCA CCC GAG ATT GAG CA-3′ and reverse 5′-TAG TAG CGA CGG GCG GTG TG-3′.

Briefly, cells were lysed in 2X loading lysis buffer [50 mM Tris-HCl (pH 6.8), 2% sodium dodecyl sulfate, 10% β-mercaptoethanol, 10% glycerol and 0.002% bromophenol blue]. The concentration of total protein was then determined by Enhanced BCA Protein Assay kit (Beyotime Institute of Biotechnology). Total proteins (25 µg/µl) from each group were then separated by 12% SDS-PAGE and transferred onto Hybrid-polyvinylidene fluoride membranes (EMD Millipore). The membranes were blocked with 5% (w/v) non-fat dried milk in TBST and washed four times with TBST at room temperature (15 min per wash) prior to incubation with the primary antibodies (all final dilution 1:1,000) in 5% non-fat dried milk in TBST) at 37°C for 1 h. The primary antibodies were: Mouse anti-CD31 (clone P2B1; cat. no. ab24590; Abcam), rabbit anti-CD146 (clone EPR3208; cat. no. ab75769; Abcam), mouse anti-FVIII (clone GMA-012; ab78852; Abcam), rabbit anti-vascular endothelial growth factor receptor 1 (VEGFR1; clone Y103; cat. no. ab32152; Abcam), rabbit anti-PDGFRA (clone D1E1E; cat. no. 3174; Cell Signalling Technology, Inc.), rabbit anti-PDGFRB (clone 28E1; cat. no. 3169, Cell Signalling Technology, Inc.), rabbit anti-argonaute 2 (Ago2; clone C34C6; cat no. 2897; Cell Signalling Technology, Inc.), Notch Activated Targets Antibody Sampler kit (cat. no. 68309; Cell Signalling Technology, Inc.), and rabbit anti-GAPDH antibody (clone EPR16891; cat. no. ab181602; Abcam). After washing, the membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse secondary antibody (1:1,000; cat. nos. ab97051 and ab6789; Abcam) at 37°C for 1 h. After washing four times with TBST at room temperature (15 min per wash), the immunoreactive protein bands were visualized by enhanced chemiluminescence (ECL) using an ECL kit (PerkinElmer, Inc.).

In a 6-well plate, a total of 30 24-h post-fertilization zebrafish embryos of the fli1a-EGFP/casper strain (Shanghai Research Centre for Model Organisms, Shanghai, China) were added to each well. DMSO (0.1%) was added to the E3 embryonic culture medium (0.58 g/l NaCl, 0.27 g/l KCl, 0.97 g/l CaCl₂·2H₂O, 0.16 g/l MgCl₂·6H₂O, 1% methylene blue; pH 7.2; all from Sigma-Aldrich; Merck KGaA) in the control group, and 31.8 mM anisomycin was used in the experimental group. After 8 h of treatment, the medium was replaced with fresh embryonic culture medium, and the phenotypic changes of zebrafish embryos were photographed under a Nikon SMZ 1500 stereomicroscope (Nikon Corporation; magnification, ×40). The numbers of fully-developed intersegmental blood vessels (ISV) and dorsal longitudinal anastomotic vessels (DLAV) in each embryo were counted. The inhibition rate of angiogenesis was calculated using the following formula (24): % inhibition of angiogenesis = [1-(average number of ISV in the embryoid body in the experimental group/average numbers of ISV and DLAV in the embryoid body in the control group)] ×100.

Total RNA samples from each group of cells were labeled using Agilent's Low RNA Input Fluorescent Linear Amplification kit (Agilent Technologies, Inc.), according to the manufacturer's protocol. Cy3-dCTP or Cy5-dCTP was incorporated during reverse transcription of 5 µg total RNA into cDNA. Different fluorescently labeled cDNA probes were mixed in 30 µl hybridization buffer (3X SSC, 0.2% SDS, 5X Denhardt's solution and 25% formamide) and applied to the microarray (CapitalBio human mRNA microarray V2.0; CapitalBio Technology, Inc.) at 42°Cfor 16 h. After hybridization, the slide was washed with 0.2% SDS/2X SSC at 42°C for 5 min, and then washed with 0.2X SSC at room temperature for 5 min. The fluorescent images of the hybridized microarray were scanned with an Agilent Whole Human Genome 4×44 microarray scanner system (Agilent Technologies, Inc.). Images and quantitative data of the gene expression levels were analyzed using the Agilent's Feature Extraction (FE) software, version 9.5 (Agilent Technologies, Inc.). Raw data were normalized with a robust multi-array average algorithm and analyzed using GeneSpring software (version 13.1; Agilent Technologies, Inc.). Differentially expressed genes (DEGs) were then identified via the fold-change (Sample group/Control group) in expression, as well as the P-value calculated from a t-test. The threshold set for upregulated and downregulated genes was: Log10 (Sample group/Control group) >3 was considered significantly upregulated expression; and log10 (Sample group/Control group) <-2 was considered significantly downregulated expression. Hierarchical clustering was then performed to demonstrate the expression profile of the DEGs among samples using MultiExperiment Viewer (version 4.9.0; http://mev.tm4.org/). Gene Ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG; https://www.genome.jp/kegg/) pathway analyses of the identified DEGs were performed using the Database for Annotation, Visualization and Integrated Discovery (DAVID; http://david.abcc.ncifcrf.gov/) database. Finally, protein-protein interaction networks were constructed using the Search Tool for the Retrieval of Interacting Genes (STRING version 11.0; https://string-db.org/) database and visualized using Cytoscape software (version 3.4.0; https://cytoscape.org/).

All steps of the luciferase reporter assay were performed as previously described (21). NIH-3T3 cells were purchased from The Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China), and cultured with DMEM medium (Gibco; Thermo Fisher Scientific, Inc.) containing 10 % FBS (Gibco; Thermo Fisher Scientific, Inc.). NIH-3T3 cells were seeded at 3×10⁴/well in 48-well plates and co-transfected with 400 ng of miR-421 mimics or miR-134-mut control, 20 ng of pGL3 cm-PDGFRA-3UTR-WT or pGL3 cm-PDGFRA-3UTR-Mut (NovoBio, Ltd.), pGL3 cm-Meg3-WT (NovoBio, Ltd.) or pGL3 cm-Meg3-Mut (NovoBio, Ltd.), and pRL-TK (Promega Corporation) using Lipofectamine 2000 reagent (Thermo Fisher Scientific, Inc.), according to the manufacturer's protocol. At 48 h post-transfection, the Firefly and Renilla luciferase activities were measured using the dual-luciferase reporter assay system (Promega Corporation).

RIP experiments were performed using the Magna RIP RNA-Binding Protein Immunoprecipitation kit (EMD Millipore). All the steps of RIP were performed as previously described (12). Briefly, cells from all groups were lysed (500 µl per plate) in a modified cell lysis buffer used for western blotting and immunoprecipitation (20 mM Tris pH7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, sodium pyrophosphate, β-glycerophosphate, Na₃VO₄ and leupeptin; Beyotime institute of Biotechnology). After lysis, each sample was centrifuged at 4°C, 1,200 × g for 10 min to clear the insoluble debris and the supernatant was pre-incubated with 20 µg protein A agarose beads (Beyotime institute of Biotechnology) by rocking for 30 min at 4°C, followed by centrifugation (4°C, 450 × g for 2 min) and transfer to a new 1.5 ml tube. The mouse anti-human Ago2 monoclonal antibody (clone C34C6; cat no. 2897; Cell Signaling Technology, Inc.; 1:100) was added for 90 min, before the re-addition of 20 µg of protein A agarose beads to capture the immune complexes. The agarose beads were then washed three times with ice-cold homogenization buffer. Then, the co-precipitated RNAs were isolated by resuspending the beads in TRIzol RNA extraction reagent (Thermo Fisher Scientific, Inc.), and extracted using the RNeasy Mini kit (Qiagen GmbH). Total RNAs were subjected to reverse transcription using a ReverTra Ace-α First Strand cDNA Synthesis kit (Toyobo Life Science). PCR amplification was performed for 31 cycles as follows: Denaturation at 95°C for 30 sec, annealing at 65°C for 30 sec and extension at 72°C for 42 sec, using Easy-Load™ PCR Master Mix (cat. no. D7251-1 ml; Beyotime Institute of Biotechnology). The amplification products were visualized by 1.2% agarose gel electrophoresis. The primers for amplification of each gene were as follows: miR-421, forward 5′-ATC AAC AGA CAT TAA TTG GGC GC-3′ and reverse 5′-GCT GTC AAC GAT ACG CTA CCT A-3′; and lncRNA-Meg3, forward 5′-GCC CAT CTA CAC CTC AC-3′ and reverse 5′-ATC CTT TGC CAT CCT G-3′.

Each experiment was performed as least three times, and data are presented as the mean ± standard error where applicable. Differences were evaluated using one-way ANOVA followed by Dunnett's test. Statistical analysis was performed by GraphPad Prism Version 5.00 (GraphPad Software, Inc.). P<0.05 was considered to indicate a statistically significant difference.

First, the association between HuOCSCs and tumour angiogenesis was investigated. Flow cytometry analysis revealed a large difference in the number of CD44⁺/CD133⁺ HuOCSCs in serous ovarian cancer tissues from four different patients (Fig. 1). Subsequently, H&E staining revealed that there were more new blood vessels in tissues from P2 patient with increased numbers of CD44⁺/CD133⁺ cells than in tissues from P4 patient with less CD44⁺/CD133⁺ cells (Fig. 2). Immunohistochemical staining confirmed that the expression of CD31 and CD34 was significantly higher in tissues from P2 patient with more CD44⁺/CD133⁺ cells than in tissues from P4 patient with less CD44⁺/CD133⁺ cells (Fig. 2). Therefore, these results indicated that higher content of CD44⁺/CD133⁺ cells in ovarian cancer tissues may be associated with increased angiogenesis.

The effect of different concentrations (1-100 mM) of anisomycin was next analysed on the proliferation of CD44⁺/CD133⁺ HuOCSCs in vitro. The MTT assay results demonstrated that after 36 h of cell treatment, the inhibition rate of cell proliferation increased significantly with the increase of anisomycin concentration (Fig. 3A). The results showed that anisomycin caused inhibition of cell proliferation in a dose-dependent manner (IC₅₀=31.8 mM). Flow cytometry analysis of Annexin V/PI double staining indicated that treatment of HuOCSCs with anisomycin for 36 h significantly increased the proportion of apoptotic cells in HuOCSCs, and significantly decreased the proportion of viable cells (Fig. 3B and C). In addition, Transwell invasion assays demonstrated that treatment of HuOCSCs with anisomycin for 36 h significantly weakened their invasive ability in Matrigel (Fig. 4A).

Next, the effects of anisomycin on angiogenesis were investigated. The results of an in vitro HUVEC capillary tubule formation assay showed that 36 h of treatment with anisomycin at the IC₅₀ dose significantly inhibited the ability of HUVECs to form 3-dimensional tubular structures in Matrigel (Fig. 4B and C). Furthermore, zebrafish embryos of the fli1a-EGFP/casper strain were incubated with anisomycin at the IC₅₀ dose for 24 h and the results demonstrated that anisomycin significantly inhibited the growth and development of ISV and DLAV (Fig. 4D and E); the inhibition rate of vascular development in the anisomycin-treated group was significantly higher compared with the DMSO-treated and non-treated groups (Fig. 4D). These results indicated that anisomycin significantly inhibited the proliferation and invasion of HuOCSCs, as well as inhibited angiogenesis of HUVECs.

First, CD44⁺/CD133⁺ HuOCSCs were divided into two groups, one group was treated with anisomycin at IC₅₀ dose, and the other group was treated with an equal volume of DMSO. After the two groups of cells were treated for 36 h as aforementioned, an equal number of cells was used to subcutaneously inoculate mice at the shoulder or hip. The two groups of nude mice were kept under the same conditions for 35 weeks. It was observed that tumours grew on both groups of nude mice at week 21 post-inoculation, but the tumours formed by DMSO-treated cells were markedly larger than those formed by anisomycin-treated cells (Fig. 5A). Over time, the volume of tumours formed by DMSO-treated HuOCSCs increased significantly, while the volume of tumours formed by anisomycin-treated HuOCSCs increased slowly; a statistically significant difference in tumour volume was observed between the two groups (Fig. 5B). At week 35, all nude mice were sacrificed, and the weights of tumours in each group were measured. The results demonstrated that the weight of the tumours formed by anisomycin-treated HuOCSCs was significantly decreased compared with that formed by DMSO-treated cells (Fig. 5C). Subsequently, H&E staining of the xenograft tumours revealed minimal neovascularization in the tumour tissues formed by anisomycin-treated HuOCSCs, whereas marked neovascularization was observed in the tumour tissues formed by DMSO-treated cells (Fig. 5D). In addition, immunofluorescence staining revealed that the density of CD31/Flt-1 and FVIII/CD146 double-positive cells in the tumour tissues formed by anisomycin-treated HuOCSCs was significantly decreased compared with that in the tumour tissues formed by DMSO-treated cells (Fig. 5E). Therefore, these results indicated that anisomycin significantly inhibited the proliferation and angiogenic ability of CD44⁺/CD133⁺ HuOCSCs in vivo.

Using cDNA microarrays, changes in the expression profiles of a total of 14,740 genes were identified between the anisomycin-treated and the untreated groups (Fig. 6A). Among them, the mRNA expression levels of 1,797 genes changed significantly following treatment with anisomycin in CD44⁺/CD133⁺ HuOCSCs. The results indicated that the expression of vascular endothelial cell markers, platelet-derived growth factors (PDGFs), Notch pathway components, and 27 tumour angiogenesis-related genes, all showed significant downregulation (Fig. 6B; Table II). GO analysis revealed that in terms of cellular components, the most significant differences in gene expression were observed for genes included in the 'integral to membrane' section (Fig. 6C). In terms of molecular function, the most significant differences in gene expression were observed for genes included in the 'receptor activity' section (Fig. 6C). In terms of biological process, the most significant differences in gene expression were observed for genes included in the 'signal transduction' section (Fig. 6C). In addition, with regards to KEGG analysis, the most significant differences in gene expression between the anisomycin-treated and untreated groups were observed for genes included in the 'pathways in cancer' and 'cytokine-cytokine receptor interaction' sections (Fig. 6C). Further analysis with the STRING bioinformatics tool (https://string-db.org/) identified a network of interactions among angiogenesis-related proteins. The software predictions indicated that the vascular endothelial cell marker, PDGFs, Notch pathway components, and the proteins encoded by 27 angiogenesis-related genes were intrinsically interrelated, with multiple predicted interactions between them (Fig. 6D). These data indicated that anisomycin significantly inhibited the expression of multiple angiogenesis-related genes in ovarian cancer stem cells.

Subsequently, the expression levels of the Notch pathway components and PDGF receptors were examined by qPCR and western blotting. qPCR results demonstrated that the mRNA expression levels of key factors of the Notch pathway (Notch-1, Dll1, Hes1 and RBPJ) and PDGFRA were significantly increased in ovarian cancer tissues from P2 patient with more CD44⁺/CD133⁺ HuOCSCs compared with those from P4 patient with less CD44⁺/CD133⁺ HuOCSCs (Fig. 7A). The results from the western blot analysis were similar those of qPCR (Fig. 7C). However, after CD44⁺/CD133⁺ HuOCSCs were treated with anisomycin, the results of qPCR and western blot analyses revealed that the expression levels of key factors of the Notch pathway (Notch-1, Dll1, Hes1 and RBPJ) and PDGFRA and PDGFRB in the anisomycin-treated cells were significantly decreased compared with those in the DMSO-treated and non-treated groups (Fig. 7B and D). In addition, immunohistochemical staining indicated that, firstly, PDGFRA was highly expressed in ovarian cancer tissues with more CD44⁺/CD133⁺HuOCSCs compared with tissues with less CD44⁺/CD133⁺ HuOCSCs (Fig. 8A); PDGFRB was mostly undetectable in both groups (Fig. 8A). Secondly, in the xenograft tumours of nude mice, the expression levels of key factors of the Notch pathway and PDGFRA were significantly increased in the tumours formed by anisomycin-treated HuOCSCs compared with those in tumours formed by DMSO-treated cells (Fig. 8B). The present results suggested that anisomycin significantly inhibited the activation of the Notch pathway and PDGFRs in HuOCSCs.

Through in-depth analysis of the cDNA microarray data, it was revealed that the expression levels of lncRNA-Meg3 in the anisomycin-treated group were significantly lower compared with that in the control group (Fig. 9A). Furthermore, qPCR results suggested an opposite trend in expression between lncRNA-Meg3 and miR-421 expression levels in both tumour samples of patients and subcutaneous xenograft tumour tissues of nude mice (Fig. 9B and C). In addition, after HuOCSCs were treated with anisomycin, their endogenous miR-421 expression levels were significantly higher compared with that of control cells (Fig. 9D).

Next, the effect of miR-421 overexpression was investigated in HuOCSCs. First, RT-qPCR results confirmed that miR-421 was successfully overexpressed following mimics transfection, compared with a mutant control or untransfected cells (Fig. S1). Following exogenous miR-421 overexpression in HuOCSCs, RT-qPCR results indicated that both expression levels of lncRNA-Meg3 and PDGFRA were significantly reduced compared with control-transfected cells (Fig. S1). According to bioinformatics analysis using TargetScanHuman (Version 7.0; http://www.targetscan.org), the mature miR-421 complementarily pairs with 7 bases at a specific site (97-104 bp) of lncRNA-Meg3, suggesting that lncRNA-Meg3 may be one of the targets of miR-421 (Fig. 9E). At the same time, it was also identified that the mature miR-421 complementarily pairs with 7 bases at a specific site (471-477 bp) of the 3′UTR of the PDGFRA gene, suggesting that PDGFRA may also be one of the targets of miR-421 (Fig. 9E). To examine the potential regulatory effect of miR-421 on lncRNA-Meg3 and PDGFRA, luciferase reporter plasmids were constructed for these two genes (Fig. 9F). The results of the luciferase reporter assay demonstrated that when miR-421 was overexpressed in cells, the luciferase activity of the PDGFRA 3′UTR reporter was significantly decreased, while the other plasmid combinations did not affect luciferase activity (Fig. 9H). In the aforementioned system, the luciferase activity did not decrease significantly after overexpression of lncRNA-Meg3 (Fig. 9I). However, when miR-421 was overexpressed in cells, expression of the luciferase gene carrying wild-type Meg3 was significantly decreased, while the other plasmid combinations did not affect luciferase activity (Fig. 9J). Finally, the results of RIP assay indicated that binding between Meg3 and Ago2 decreased significantly following treatment of HuOCSCs with anisomycin; however, there was no significant change in the binding level between miR-421 and Ago2 (Fig. 9G and K). The present results suggested that lncRNA-Meg3 and PDGFRA may be potential targets of miR-421; anisomycin downregulated the expression of endogenous lncRNA-Meg3, leading to redundant miR-421, which subsequently inhibited PDGFRA expression.

MTT assay results showed that treating miR-421-overexpressing HuOCSCs with anisomycin for 24 h resulted in a significant inhibition of cell proliferation; the inhibition rate of proliferation was significantly higher compared with that of HuOCSCs overex-pressing miR-421 alone or HuOCSCs treated with anisomycin alone (Fig. 10A). qPCR results demonstrated that, following treatment of miR-421-overexpressing HuOCSCs with aniso-mycin for 24 h, the mRNA expression levels of endogenous lncRNA-Meg3, PDGFRA, Notch1, HES1 and RBPJ were significantly decreased; the expression levels were significantly lower than those of HuOCSCs overexpressing miR-421 alone or HuOCSCs treated with anisomycin alone (Fig. 10B). These results suggested that overexpression of miR-421 significantly enhanced the inhibitory effect of anisomycin on HuOCSCs by suppressing the activity of the Notch-1 pathway.