Exosomal miR-99a-5p is elevated in sera of ovarian cancer patients and promotes cancer cell invasion by increasing fibronectin and vitronectin expression in neighboring peritoneal mesothelial cells

Dulbecco’s modified Eagle’s medium (DMEM) and Roswell Park Memorial Institute (RPMI) 1640 medium were obtained from Nacalai Tesque (Kyoto, Japan). Fetal bovine serum (FBS; #172012) was purchased from Sigma Aldrich (St. Louis, MO). Antibody against CD63 (#11–343-C025) was purchased from EXBIO (Prague, Czech Republic). Donkey anti-mouse IgG 10 nm gold (#ab39593) was purchased from Abcam (Cambridge, UK). Growth factor-reduced basement membrane protein (Matrigel; #356230) was purchased from Corning (New York, NY). Antibodies against vitronectin 65/75 (D-8; #sc-74,484) and fibronectin (EP5; #sc-8422) were purchased from Santa Cruz Biotechnology (Dallas, TX). Antibody against β-actin (#4967) was purchased from Cell Signaling Technology (Danvers, MA).

The HeyA8 cell line was generously provided by Dr. Anil Sood (MD Anderson Cancer Center, Houston, TX) in 2010 and the TYK-nu cell line was purchased from Health Science Research Resources Bank (JCRB0234.0; Osaka, Japan) in 2011. Cells were authenticated by short tandem repeat DNA profiling at Takara-Bio Inc. (Otsu, Japan) and were used for this study within 6 months of resuscitation. The IOSE cell line was generously provided by Dr. Masaki Mandai (Kyoto University, Kyoto, Japan) in 2013. Briefly, ovarian surface epithelial cells were collected from normal ovaries and transfected with SV40 large T antigen and human telomerase reverse transcriptase [9]. HPMCs were isolated from the omentum of patients undergoing surgery at Osaka University Hospital as previously described [10]. Written informed consent was obtained from each patient before surgery. Briefly, small pieces of omentum were trypsinized at 37 °C for 30 min and the cell-containing suspension was filtered through a 200-nm pore nylon mesh and centrifuged at 500×g for 5 min. The cells were cultured in RPMI 1640 supplemented with 20% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin and incubated at 5% CO₂ and saturated humidity at 37 °C. The cells were harvested during the second or third passage after primary culture for experiments. Mycoplasma contamination had been routinely checked using EZ-PCR Mycoplasma Test Kit (Biological Industries, Kibbutz Beit Haemek, Israel).

Conditioned medium (CM) containing exosome-depleted FBS (prepared by overnight ultracentrifugation at 100,000×g at 4 °C) was prepared by incubating cells grown at subconfluence for 48 h. CM was centrifuged at 2000×g for 10 min at 4 °C and the supernatant fraction was filtered through 200-nm pore size filters. The resulting cell-free medium was ultracentrifuged at 100,000×g for 70 min at 4 °C using a Beckman™ L-90 K ultracentrifuge (Brea, CA). The supernatant fraction was discarded, and then the exosome-containing pellet was resuspended in phosphate-buffered saline (PBS) and ultracentrifuged under the same conditions. The pellet was finally resuspended in PBS and the amount of exosomal protein was assessed by the Lowry method (Bio-Rad, Hercules, CA).

Electron microscopy was performed as described using a transmission electron microscope (H-7650; Hitachi, Ltd., Tokyo, Japan).

Exosome suspensions were diluted 1000-fold with PBS and nanoparticle tracking analysis was carried out using a NanoSight LM10V-HS particle analyzer (Malvern Instruments Ltd., Worcestershire, UK).

Total RNA was extracted using TRIzol reagent (#15596–018; Life Technologies, Carlsbad, CA:). RNA isolated from cells and exosomes was analyzed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc. Santa Clara, CA).

miRNA microarrays using the GeneChip miRNA 4.0 Array (Affymetrix, Santa Clara, CA) were performed and analyzed by Filgen (Nagoya, Japan). Briefly, 1000-ng miRNA samples were biotin-labeled using a Flash Tag_TM Biotin HSR RNA Labeling Kit for Affymetrix GeneChip miRNA arrays (Affymetrix) according to the manufacturer’s protocol. Hybridization solution was prepared using 110.5 μL hybridization master mix and 21.5 μL biotin-labeled sample. The array was incubated using the GeneChip Hybridization Oven 645 (Affymetrix) and washed using the GeneChip Fluidics Station 450 (Affymetrix) according to the manufacturer’s protocol. The washed array was analyzed using the GeneChip Scanner 3000 7G (Affymetrix).

miRNA qRT-PCR was performed using the StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA). Total RNA was transcribed into cDNA using the TaqMan MicroRNA Reverse Transcription Kit (#4366596; Applied Biosystems). Mature miR-99a-5p was assayed using the TaqMan assay (#A25576; hsa-miR-99a-5p). To normalize miRNA expression levels, cel-miR-39 (#4427975; Applied Biosystems) was used as an exogenous control for serum miRNA, and RNU6B (Applied Biosystems; #001093) was used as an endogenous control for cellular miRNA. Each qRT-PCR assay was performed in triplicate, and the relative expression levels of miR-99a-5p were calculated using the 2^-∆∆Ct method.

Blood samples were collected from healthy volunteers (n = 20), patients with benign ovarian tumors (n = 26), and ovarian cancer patients (n = 62) with approval from the Institutional Review Board of Osaka University. Written informed consent was obtained from every participant before sample collection. From 26 ovarian cancer patients, blood samples were collected twice; once at admission for primary debulking surgery (PDS; 2 or 3 days before surgery) and once at admission for initial postoperative chemotherapy (36 ± 12 days after surgery). Written informed consent was obtained from every participant for the use of their samples. Blood samples were centrifuged at 1500×g for 10 min at 4 °C and the upper sera phases were transferred to new tubes and stored at − 80 °C until further use.

miRNA was extracted from 200-μL serum samples using the miRNeasy Serum/Plasma Kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol.

In miRNA experiments, miRNAs were used to transfect HPMCs after reaching 60–70% confluence. After 24 h, HPMCs were harvested and plated onto 25-μg Matrigel-coated 24-well cell culture inserts with 8-μm pore membranes (#353097; Corning). After HPMCs reached confluence, 5 × 10⁴ ovarian cancer cells were plated onto the HPMC monolayer with 0.1% bovine serum albumin (BSA)/DMEM and allowed to invade for 24 h. Cell culture media consisting of 10% FBS/DMEM was placed in the lower chamber as a chemoattractant. Once the HPMCs reached confluence in exosome experiments, the culture media was changed to 0.1% BSA/DMEM with 100 μg/mL exosomes 24 h before treatment with cancer cells. Noninvading cells were removed with a cotton swab and invading cells that migrated to the trans-side of the membrane were fixed with methanol, stained with Giemsa, and counted using the image analysis application of the BZ-X700 microscope (Keyence, Osaka, Japan).

HPMCs were transfected with precursor miRNA (hsa-miR-99a-5p: #AM17100) or inhibitor miRNA (anti-miR-99a-5p; #AM17000) at a concentration of 30 nM. Pre-miR miRNA precursor negative control #1 (#AM17110) was used as a control. All oligonucleotides were purchased from Thermo Fisher Scientific (Waltham, MA). Oligonucleotide transfection was performed using Lipofectamine 3000 (#L3000–008; Thermo Fisher Scientific) according to manufacturer’s instructions. One day after transfection (24 h), cells were collected for subsequent procedures.

HPMCs were transfected with miR-99a-5p or negative control miRNA for 24 h and lysed with a buffer composed of 50 mM Tris-HCl pH 7.5, 2% sodium deoxycholate, and protease inhibitor (Nacalai Tesque). Protein extracts were reduced, alkylated, and digested overnight. Samples were labeled with TMT 6-plex reagents (Thermo Fisher Scientific) and then mixed before sample fractionation and clean-up. Labeled samples were analyzed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) using a Q Exactive Mass Spectrometer (Thermo Fisher Scientific) equipped with an UltiMate 3000 Nano LC System (Thermo Fisher Scientific). Raw data were processed by Mascot v2.1 (Matrix Science, London, UK). Protein expression levels were compared between the two samples and pathway annotation for differentially expressed proteins was performed using DAVID 6.8 analysis (https://​david.​ncifcrf.​gov).

A total of 5 × 10⁵ cells were plated onto 6-well plates and lysed with radioimmunoprecipitation assay buffer (RIPA buffer; 25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% sodium deoxycholate, 1% NP-40, and 0.1% sodium dodecyl sulfate). Lysates were separated by 5–20% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes, followed by incubation with primary antibodies (vitronectin, 1:1000 in 5% BSA; fibronectin, 1:1000 in 5% BSA; β-actin, 1:2000 in 5% BSA), and then incubation with the corresponding secondary horseradish peroxidase–conjugated IgG. The proteins were visualized with an electrochemiluminescent system (PerkinElmer Life Science, Waltham, MA).

Statistical analysis was performed using R version 3.2.3 and JMP version 13.0.0 (SAS Institute Japan Ltd., Tokyo, Japan). Unless otherwise stated, the data are presented as the mean ± standard deviation (SD), and statistical significance was determined by a Student’s t-test. Multiple comparison analysis was performed by analysis of variance followed by Holm’s step-down method. In graphs showing relative expression levels of miR-99a-5p in patient sera, the box-and-whisker plots indicate median, interquartile range, and range; statistical significance was determined by Wilcoxon’s rank sum test. For multiple comparisons, statistical significance was determined by the Kruskal-Wallis test followed by Holm’s step-down method. Wilcoxon’s signed rank test was used to assess differences in serum miR-99a-5p expression levels between patients before and after surgery. Differences were considered statistically significant at p < 0.05. A receiver operating characteristic (ROC) curve was generated and the area under the ROC curve (AUC) was calculated to evaluate the diagnostic accuracy of miR-99a-5p.

First, we isolated EOC-derived exosomes from cell culture media using multi-step centrifugation. Two EOC cell lines, namely HeyA8 and TYK-nu, were used to isolate exosomes, as these cell lines are derived from high-grade serous carcinomas. An immortalized normal ovarian epithelial cell line, IOSE, was used as a control. Exosomes showed positive staining for CD63 (a representative exosome marker), exhibited a round-shaped morphology, and were approximately 100 nm in size as determined by electron microscopy (Fig. 1a) and nanoparticle analysis (Fig. 1b). To confirm whether exosomes were successfully purified, we analyzed the profile of total RNA extracted from exosomes using capillary electrophoresis. In contrast to total RNA extracted from cells, 18S and 28S ribosomal subunits were hardly detected in exosome sample extracts. Instead, the majority of total RNA in exosomes was below 2 kb, suggesting that most exosomal RNAs were small RNAs such as miRNAs, as previously reported [11] (Fig. 1c).

Next, to identify which exosomal miRNAs were highly expressed in ovarian cancer, miRNA microarrays were performed using exosomes from HeyA8, TYK-nu, and IOSE cells. Nine miRNAs were highly (> 2.0) expressed in both HeyA8 and TYK-nu cells in comparison with IOSE cells, as shown in Fig. 1d. Among those, we focused on miR-99a-5p, as miRNA qRT-PCR assays validated that its expression was significantly increased in EOC cell-derived exosomes; an increase of 3.4-fold in HeyA8 exosomes and 5.0-fold in TYK-nu exosomes compared with IOSE exosomes was observed (Fig. 1e).

To evaluate the diagnostic value of miR-99a-5p as an EOC biomarker, miR-99a-5p expression levels were measured by qRT-PCR in sera from 62 EOC patients. Sera from 26 patients with benign ovarian tumors and 20 healthy volunteers were analyzed as controls. Patient characteristics are summarized in Table 1. In patients with EOC, serum miR-99a-5p levels were significantly higher than in healthy controls and patients with benign tumors (2.8-fold and 1.7-fold, respectively; Fig. 2a). EOC consists of the following four major histological subgroups: serous, clear-cell, endometrioid, and mucinous. However, among the histological subtypes, serum miR-99a-5p expression did not significantly differ (Fig. 2a). While the relative expression levels of miR-99a-5p in stage I-II EOC patients were 2.0 (1.4–3.6), expression levels in stage III-IV patients tended to increase to 2.5 (1.8–5.4, p = 0.12), indicating that miR-99a-5p reflects tumor burden. To further analyze this, sera were collected from 26 EOC patients before (2 or 3 days before PDS) and after surgery (36 ± 12 days after PDS) and then miR-99a-5p expression levels were compared (Fig. 2b). When the average miR-99a-5p expression levels in healthy volunteers (n = 20) was set to 1.0, relative expression levels before surgery were 1.8 (1.2–4.5) and significantly declined to 1.3 (0.8–2.1) after surgery (p = 0.003), strongly suggesting that serum miR-99a-5p reflects tumor burden and is derived from ovarian cancer cells. CA125, a conventional biomarker of EOC, is expressed as a membrane-bound protein at the surface of cells that undergo metaplastic differentiation into Müllerian-type epithelium or is released in soluble form in bodily fluids [12]. Thus, serum miR-99a-5p expression levels were compared with CA125 values and a positive correlation was observed in patients with EOC (r = 0.48, p < 0.0001; Fig. 2c).

To further evaluate the possibility of serum miR-99a-5p as a biomarker of EOC, ROC analysis for discriminating EOC from healthy volunteers was performed (Fig. 3a). When the cut-off for miR-99a-5p was set to 1.41, the sensitivity was 0.85 and the specificity was 0.75 (AUC = 0.88). For comparison with CA125, a ROC curve for discriminating EOC from healthy volunteers was similarly calculated. When the cut-off for CA125 was set to 35.0 U/mL, the sensitivity and specificity of serum CA125 were 0.82 and 0.95, respectively (AUC = 0.91). There was no significant difference between the AUCs of miR-99a-5p and CA125. When CA125 was combined with miR-99a-5p, the AUC improved from 0.91 to 0.95 (p = 0.09; Fig. 3a). Further statistical analyses were performed to develop classification models that can distinguish among different EOC subtypes. As shown in Additional file 1: Figure S1, the miR-99a-5p classification model had AUC, sensitivity, and specificity values of 0.60, 0.84, and 0.40, respectively, for serous EOC; 0.54, 0.33, and 0.91, respectively, for clear-cell EOC; 0.68, 0.67, and 0.82, respectively, for endometrioid EOC; and 0.56, 0.33, and 0.91, respectively, for mucinous EOC. Unfortunately, this model failed to distinguish between the different EOC subtypes.

Detecting EOC patients at an early stage to improve prognosis is clinically indispensable. For this reason, ROC analysis for discriminating early-stage (stage I-II) EOC from healthy volunteers was performed (Fig. 3b). When the cut-off of miR-99a-5p was set to 1.36, the sensitivity was 0.90 and the specificity was 0.75 (AUC = 0.85). The sensitivity and specificity of serum CA125 were 0.68 and 0.95, respectively (AUC = 0.84); Therefore, there was no significant difference between the AUCs of miR-99a-5p and CA125. When CA125 was combined with miR-99a-5p, the AUC improved from 0.84 to 0.91 (p = 0.13; Fig. 3b). Finding a noninvasive method to distinguish benign tumors from malignant cancers is critical as it is currently impossible to establish a definite diagnosis without surgery. For this reason, ROC analysis for discriminating EOC from benign tumors was performed (Fig. 3c). When the cut-off of miR-99a-5p was set to 1.36, the sensitivity was 0.87 and the specificity was 0.54 (AUC = 0.70). The sensitivity and specificity of serum CA125 were 0.73 and 0.75, respectively (AUC = 0.79) and thus, there was no significant difference between the AUCs of miR-99a-5p and CA125. When CA125 was combined with miR-99a-5p, the AUC improved from 0.79 to 0.81 (p = 0.51; Fig. 3c).

Since exosomal miR-99a-5p is secreted from ovarian cancer cells and serum miR-99a-5p reflects tumor burden, we were encouraged to analyze how exosomal miR-99a-5p affects ovarian cancer progression. During peritoneal dissemination of ovarian cancer, cells detached from the primary site of origin attach to the peritoneal cavity, which is lined with a single layer of mesothelial cells. Therefore, the initial interaction of cancer cells with mesothelial cells is a key process of ovarian cancer progression and several recent reports have shown that EOC-derived exosomes mediate this process [13, 14]. Thus, we collected HPMCs from gynecological surgeries and treated them with EOC-derived exosomes. miR-99a-5p levels increased significantly in HPMCs treated with EOC (HeyA8 and TYK-nu)-derived exosomes compared with those treated with IOSE-derived exosomes (Fig. 4a). Since TYK-nu derived exosomes drastically increased expression levels of miR-99a-5p in HPMCs (5.2-fold), TYK-nu derived exosomes were used for subsequent experiments. The role of EOC-derived exosomes in ovarian cancer invasion was studied using an in vitro invasion assay. To mimic the mesothelial layer covering the omentum and other peritoneal organs, a monolayer of HPMCs was plated onto growth factor-reduced Matrigel (Fig. 4b). Pre-treatment of HPMCs with TYK-nu-derived exosomes at a concentration of 100 μg/mL significantly stimulated ovarian cancer cell invasions (Fig. 4c). To clarify the role of exosomal miR-99a-5p in ovarian cancer progression, HPMCs were transfected with miR-99a-5p (Fig. 4d) and an in vitro invasion assay was performed. While the enforced expression of miR-99a-5p did not affect the proliferation of HPMCs (Fig. 4e), it significantly promoted EOC cell invasion (Fig. 4f). Furthermore, miR-99a-5p expression in HPMCs was inhibited by transfection with miR-99a-5p specific antagonist (Fig. 4g). An in vitro invasion assay was subsequently performed and the inhibition of miR-99a-5p expression in HPMCs also did not affect the proliferation of HPMCs (Fig. 4h), but significantly attenuated EOC cell invasion (Fig. 4i). Collectively, these results show that the exosomal transfer of miR-99a-5p into HPMCs promotes EOC cell invasion without affecting cell viability.

To gain insights into the mechanism through which exosomal miR-99a-5p promotes EOC cell invasion, comprehensive proteomics analyses were performed through the TMT method. In HPMCs transfected with miR-99a-5p, several proteins were upregulated (more than 1.5-fold) compared with those transfected with negative control miRNA (Table 2). Since the enforced expression of miR-99a-5p in HPMCs promoted EOC cell invasion, we focused on fibronectin and vitronectin, which are related to extracellular matrix organization. As a validation assay, western blotting was performed and HPMCs transfected with miR-99a-5p showed significantly enhanced expression of fibronectin and vitronectin compared with that in cells transfected with negative control (Fig. 5a). Similarly, transfection of HPMCs with EOC-derived exosomes, compared with IOSE-derived exosomes, drastically enhanced the expression of fibronectin and vitronectin (Fig. 5b). The enhanced expression of these proteins was significantly attenuated by transfection with the specific inhibitor of miR-99a-5p (Fig. 5c, d), suggesting that EOC-derived exosomes upregulate the expression of fibronectin and vitronectin in HPMCs through the transfer of miR-99a-5p.