The present study was approved by the Medical Ethics Committee of the Second Affiliated Hospital of Zhengzhou University (Henan, China). A total of 10 patients with peritoneal metastatic epithelial ovarian cancer (EOC group) and 10 subjects without epithelial ovarian cancer with acute pelvic peritonitis (NC group) were enrolled in the present study. The clinical data of the participants are described in Table I. All participants were recruited at The Second Affiliated Hospital of Zhengzhou University (Zhengzhou, China), and written informed consent was obtained from each participant prior to enrolment. Peritoneal exosomes were isolated from a sample of ascites from each patient with EOC or a peritoneal lavage sample from subjects in the NC group using ExoQuick^™ exosome precipitation solution (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) following the manufacturer's instructions. Briefly, each sample of ascites or peritoneal lavage specimen was pre-cleared by centrifugation at 3,000 × g for 15 min under 4°C to remove cells or cell debris followed by mixing with precipitation solution on a 4:1 ratio (v/v). The mixture was then refrigerated under 4°C for 12 h, after which the exosomes in the mixture were pelleted by centrifugation at 1,500 × g for 30 min under 4°C. The exosome pellet extracted from each sample was then re-suspended in 200 µl PBS, validated by a transmission electron microscopy and stored under 4°C for ≤1 h before further analysis. A total of 30 µl supernatant from exosome isolation and 20 µl of each exosome sample suspended in PBS were subjected to western blotting using a mouse monoclonal antibody against human CD63 (dilution of 1:1,000) (catalog no. ab213090; Abcam, Cambridge, UK) to evaluate the efficiency of exosome precipitation. In western blotting, the ProteoPrep^® Total Extraction Sample kit (Sigma-Aldrich, Merck KGaA Darmstadt, Germany) was to separate proteins from exosomes according to the manufacturer's protocol. Samples were normalized for protein concentration by a BCA assay. The volumes (20 µg) of each sample were loaded on a acrylamide-bisacrylamide (10 or 12.5%) gel. Following electrophoresis, the protein was transfected with PVDF membrane. Skimmed milk powder (5%) blocked PVDF for 3 h at room temperature. The secondary antibody (Goat Anti-mouse IgG H&L, catalog no. ab6785, Abcam, Cambridge, UK) has a dilution of 1:5,000, incubated in TBS-T buffer (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) for 1 h at room temperature. Proteins were visualized using an enhanced chemiluminescence solution (Life Sciences; Thermo Fisher Scientific, Inc.).

A total of 2 sequencing libraries of peritoneal exosomal miRNAs from EOC- or NC-group specimens were constructed using a sequencing library construction kit provided by Gene Denovo (Guangzhou, China) following manufacturer's instructions. Briefly, 10 samples of PBS-suspended exosomes (80 µl) from participants in each group were mixed, and total RNA was extracted from the mixture using TRIzol. Small RNAs with 18–30 nucleotides were separated by SDS-PAGE (15%). After the ligation of 5′ and 3′ adapters to both ends of small RNAs by nested-polymerase chain reaction (PCR), the ligation products were reversely transcribed and amplified by PCR. The products (size, 140–160 bp) were further separated by agarose gel (3.5%) to generate the cDNA library. A total of 2 cDNA libraries generated from EOC or NC samples were then subjected to high throughput sequencing by Gene Denovo using the Illumina HiSeqTM 2500 system (Illumina, Inc., San Diego, CA, USA).

To obtain small RNA clean reads, raw reads data were first filtered using the steps as follows: i) Removal of low-quality reads containing >1 low-quality (quality score ≤20) base or containing unknown nucleotides (N); ii) removal of reads without 3′ or 5′ adapters; iii) Removal of reads containing 3′ and 5′ adapters but no small RNA fragments; iv) removal of reads containing ploy-A sequence within small RNA fragments and v) removal of reads that are shorter than 18 nucleotides.

All clean reads were then aligned to human small RNA sequence data that were obtained from GeneBank (release 209.0, http://www.ncbi.nlm.nih.gov/genbank/) and Rfam databases (version 11.0, http://rfam.xfam.org/) to filter any ribosomal RNA, small cytoplasmic RNA, small nucleolar RNA, small nuclear RNA or transfer RNA. The clean reads were also aligned to a reference human genome [Homo sapiens (assembly GRCh38.p12)] to remove reads mapped to exons or introns that might be fragments from mRNA degradation or reads mapped to repeat sequences. Filtered small RNA clean reads were then referred to as clean tags, which were mapped in miRBase database (release 21, http://www.mirbase.org/index.shtml) to annotate any known human miRNA.

The expression level of every annotated miRNA in EOC or NC sample group was first normalized to transcripts per million (TPM) using the following formula:

and A log₂ (FC) value of each annotated miRNA in NC or EOC sample group was calculated using the following formula:

where TPM_EOC and TPM_NC were TPM of a miRNA in EOC and NC samples, respectively, and all TPM=0 were adjusted to TPM=0.01 to avoid infinity error. miRNAs with log₂ (FC) >1 OR <-1 between EOC and NC samples were considered upregulated or downregulated, respectively. The annotated miRNAs were clustered according to their expression patterns to generate the heat map of miRNA expression in NC and EOC samples.

Target gene candidates of annotated miRNAs in each sample group were first predicted using RNAhybrid (version 2.1.2, http://bibiserv.cebitec.uni-bielefeld.de/rnahybrid/), miRanda (version. 3.3a, http://miranda.org.uk/) and TargetScan (version. 7.0, http://genes.mit.edu/targetscan.test/ucsc.html) databases. The genes predicted using all three databases were considered potential target genes. The preliminary miRNA pathway enrichment analysis was performed using the DIANA-miRPath (version 3, http://snf-515788.vm.okeanos.grnet.gr/) online software. DIANA-miRPath performs GO and KEGG enrichment analysis of genes in TarBase (version 7, http://diana.imis.athena-innovation.gr/DianaTools/index.php?r=tarbase/index) that have been validated as targets of given miRNAs. For Gene Ontology (GO) enrichment analysis, predicted target genes were mapped to GO terms in the Gene Ontology database (http://www.geneontology.org/). Significantly enriched GO terms were determined by an adjusted P<0.05 that was calculated by hyper-geometric distribution and false discovery rate correction. For Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, predicted miRNAs target genes were mapped to KEGG annotation in the KEGG database (http://www.genome.jp/kegg/pathway.html), and significantly enriched KEGG annotations were determined using the same method as that for Gene Ontology enrichment analysis.

The expression levels of miR-149-3p or miR-222-5p in exosomes from 10 patients with epithelial ovarian cancer (EOC group) and 10 subjects without epithelial ovarian cancer (NC group) were evaluated by RT-qPCR using a custom-made microRNA qPCR kit (GeneCopoeia, Inc., Rockville, MD, USA) following the manufacturer's instructions. miRNAs were first reverse transcribed using a reverse transcription reaction (miRNA First-Strand cDNA Synthesis kit, GeneCopoeia, Inc., Rockville, MD, USA) mix at 37°C for 60 min followed by inactivation at 85°C for 5 min. The primer sequence for has-miR-149-3p detection was 5′-GGCUCCGUGUCUUCACUCCCAAA-3′ and for has-miR-222-5p was 5′-AGUAGCCAGUGUAGAUCCUAAA-3′. qPCR reaction was performed following a standard three-step method as described in Table I.

The qPCR reaction was performed using a SimpliAmp thermocycler (Thermo Fisher Scientific, Inc.) and was monitored by a SYBR Green fluorescent dye. Semi-quantitative analysis was first performed to normalize the expression level of detected microRNAs to snRNA U6 (GeneCopoeia, Inc.) in each sample using the 2^−ΔΔCq method. The fold change (FC) of microRNA expression level in each EOC-group sample was compared to the mean expression level in the NC group samples (20).

The associations between the expression of specific miRNAs and the survival of patients with ovarian cancer were investigated by Kaplan-Meier curve (with a log-rank test) that was generated using the UCSC Xena browser (http://xena.ucsc.edu/). The sequencing data from 595 publically available samples from patients with ovarian cancer and clinical record from The Cancer Genome Atlas ovarian cancer database (https://cancergenome.nih.gov/) were analyzed. Graphpad Prism software (version 7.0; GraphPad Software, Inc., La Jolla, CA, USA) was employed for statistical analysis unless otherwise indicated. Student's t-test was used for significance test, and P<0.05 was considered statistically significant.

To characterize the differences in peritoneal exosomal miRNA expression patterns between NC participants and patients with EOC, exosomes were isolated from 10 samples of ascites from patients with ovarian cancer and 10 peritoneal lavage samples from participants without cancer. The clinical information of these patients is shown in Table II. Following total exosome precipitation as verified by western blotting and transmission electron microscopy (Fig. 1), small RNAs from two groups of exosome samples were extracted, and differentially expressed miRNA between NC and EOC sample were detected by next-generation sequencing. Subsequently, bioinformatic analysis was performed. As described in the materials and method section, annotated miRNAs with TPM_EOC two-fold higher compared with TPM_NC with log₂ (FC)>1 were considered upregulated in EOC samples compared to NC samples. Similarly, miRNAs with log₂ (FC)<-1 were determined to be downregulated. Among the 748 annotated miRNAs, 249 miRNAs were upregulated and 317 were downregulated in EOC samples compared with NC samples (Fig. 2A and B). These differentially expressed miRNAs were clustered according to changes in expression level in a heat map (Fig. 2C) to further illustrate that there are different patterns of miRNA expression between NC and EOC samples. These data indicated that miRNAs in peritoneal exosomes were differentially expressed between NC participants and patients with EOC.

miRNAs regulate the expression level of their specific target genes mostly by decreasing the efficiency of protein translation, therefore affecting different biological processes in cells. Considering that peritoneal exosomes are secretory extracellular vesicles from cells within the abdominal cavity and that exosomal contents may affect the biological processes in recipient cells when fused with exosome, changes in the expression of peritoneal exosomal miRNAs may therefore reflect or cause physiopathological changes in peritoneal cells and the associated microenvironment.

To investigate the potential effect of changes in the expression pattern of peritoneal exosomal miRNAs on EOC oncogenesis or abdominal cavity metastasis, GO enrichment and KEGG pathway enrichment analysis were performed on the predicted target genes of upregulated and downregulated miRNAs. A total of 44,955 genes were identified as target genes of all differentially expressed miRNAs. The significantly enriched GO terms of the genes targeted by upregulated and downregulated miRNAs are summarized in Tables III and IV, respectively, and further compared in Fig. 3. Notably, the genes targeted by upregulated miRNAs encrypt proteins that are mostly located on the cell membrane or in the extracellular microenvironment (EM) that is involved in signaling and cell communication, implying that the upregulation of these miRNAs might inhibit cell-to-cell or cell-to-EM communication. To get a better understanding of the effect of the change in the expression levels of peritoneal exosomal miRNAs between NC and EOC samples, KEGG pathway enrichment analysis of the target genes of upregulated and downregulated miRNAs was performed. The top 20 significantly enriched KEGG pathways are highlighted and summarized in Fig. 4. The genes that are targeted by miRNAs that were downregulated in the EOC group compared with the NC group were most significantly involved in pathways with cancer-promoting potential, such as phosphoinositide 3-kinase-Akt, Wnt or mitogen-activated protein kinase signaling pathways. These results suggested that changes in the expression of peritoneal exosomal miRNAs might reflect and have potential effects on the physiopathological features of ovarian cancer.

To further investigate the effect of changes in the expression of peritoneal exosomal miRNAs on the pathophysiology of ovarian cancer, the top 10 most significantly upregulated miRNAs with TPM_EOC >50 were sorted (Table IV). A total of three miRNAs with an increased expression that was associated with a decreased survival in patients with ovarian cancer were identified. As indicated in Fig. 5, a high expression of miR-149-3p and miR-222-5p was significantly correlated with a decrease in the 5-year and overall survival of patients with EOC as revealed by Kaplan-Meier curve analysis. A high expression of miR-1246 was also associated with a poor survival in patients with EOC, but there was no statistical significance (P>0.05). To the best of our knowledge, the effects of miR-149-3p and miR-222-5p on the pathophysiology of EOC have not been adequately characterized by other studies. Considering that these two miRNAs were significantly upregulated in EOC samples compared with the NC samples, it was inferred from these results that miR-149-3p and miR-222-5p may participate in the oncogenesis or progression of EOC.

The association between EOC and the upregulation of miR-149-3p or miR-222-5p in peritoneal exosomes was further verified by evaluating the expression level of these two miRNAs in the peritoneal exosomes that were obtained from the 10 patients with EOC and 10 NC participants. As indicated in Fig. 6, RT-qPCR results indicated that the expression level of miR-149-3p and miR-222-5p was significantly increased in EOC samples compared with the NC samples. These data suggested that miR-149-3p and miR-222-5p were associated with EOC and the miRNAs might be involved in the progression of EOC with the potential to serve as a biomarker for estimating the prognosis of patients with EOC.