{"paper_id":"4c8b8810-5f37-4b5f-bdd9-dd17ca7a4f9a","body_text":"R E S E A R C H Open Access\nComparison of benign peritoneal fluid- and\novarian cancer ascites-derived extracellular\nvesicle RNA biomarkers\nCindy M. Yamamoto 1* , Melanie L. Oakes 1, Taku Murakami 1, Michael G. Muto 2, Ross S. Berkowitz 2\nand Shu-Wing Ng 2,3\nAbstract\nBackground: Extracellular vesicles (EVs) are considered as a new class of resources for potential biomarkers. We\nanalyzed expression of specific mRNA and miRNA in EVs derived from ovarian cancer ascites and the ideal controls,\nperitoneal fluids from benign patients for potential early detection and prognostic biomarkers.\nMethods: Fluids were collected from subjects with benign cysts or endometrioma ( n = 10), or low/high grade\nserous ovarian carcinoma (n = 8). EV particles were captured using primarily ExoComplete filterplate or ultracentrifugation\nand analyzed by nanoparticle tracking analysis, ELISA, and scanning electron microscopy. EV RNAs extracted from two\nascites and three peritoneal fluids were submitted for next-generation sequencing. Thee x p r e s s i o no f3 4m R N Aa n d1 8\nmiRNAs in the EVs isolated from patient fluids and cell line media was determined using qPCR.\nResults: EVs isolated from patient samples had concentrations greater than 10 10 EV particles/mL and 30%\nwere EpCAM-positive based on ELISA. EV particle sizes averaged 113 ± 11.5 nm. The qPCR studies identified\nfive mRNA ( CA11, MEDAG, LAMA4, SPINT2, NANOG )a n ds i xm i R N A( let-7b, miR23b, miR29a, miR30d, miR205,\nmiR720) that were significantly differentially expressed between cancer ascites and peritoneal fluids. In addition, CA11\nmRNA was decreased to 0.5-fold and SPINT2 and NANOG mRNA were significantly increased up to 100-fold in\nconditioned media of cancer cells compared to immortalized ovarian surface and fallopian tube epithelial cell\nlines, the hypothesized cells of origin for ovarian cancer development.\nConclusions: This study indicates that EV mRNA profiles can reflect the disease stage and may provide a potentially\nnovel source for discovery of biomarkers in ovarian cancer.\nKeywords: Extracellular vesicles, Ovarian cancer, Biomarkers, Ascites, Peritoneal fluids\nBackground\nOvarian cancer is the fifth-leading cause of cancer deaths\nin women [ 1]. With a lack of early obvious symptoms,\nwomen are frequently diagnosed with advanced stage dis-\nease. Approximately 60% of women are diagnosed at stage\n3 or higher where the 5-year survival rate is below 30%\n[2]. In contrast, only 15% of cases are diagnosed at stage 1,\nwhen the tumor is localized to the primary site and pa-\ntients have a 5-year survival rate of about 92%. In addition,\npatients with metastatic ovarian cancer frequently\nexperience high recurrence rates within 16 – 22 months\nafter conventional platinum-based combination chemo-\ntherapy. Identification of novel, specific and sensitive bio-\nmarkers for screening, monitoring or prediction may\nimprove clinical outcomes and survival.\nScreening tests currently used for ovarian cancer de-\ntection include pelvic examination, transvaginal ultra-\nsound, and cancer antigen 125 (CA125). If an adnexal\nmass is detected by physical examination and/or ultra-\nsound, surgery is ultimately needed for the confirmed\novarian cancer diagnosis and staging. CA125 is more\nroutinely used as a marker for disease recurrence and\ntreatment response [ 3]. It has been shown to be elevated\nin 80% of epithelial ovarian carcinomas, but its increase\n* Correspondence: cyamamoto@hitachi-chemical.com\n1Hitachi Chemical Co. America, Ltd. R and D Center, 1003 Health Sciences Rd,\nIrvine, CA 92617, USA\nFull list of author information is available at the end of the article\n© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0\nInternational License ( http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and\nreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to\nthe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver\n(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 \nhttps://doi.org/10.1186/s13048-018-0391-2\n\nin other conditions such as endometrial, pancreatic and\nbreast cancer and certain benign conditions have limited\nits use as an early screening marker [ 4, 5]. In addition,\nannual screening with both CA125 and transvaginal\nultrasound has not reduced ovarian cancer mortality\ncompared with usual care [ 6]. Further research is neces-\nsary to discover and identify biomarkers that would be\neffective in early ovarian cancer screening.\nRecently, extracellular vesicles (EVs) have been analyzed\nfor their potential as ovarian cancer disease biomarkers\n[7]. EVs comprise of exosomes (30 – 100 nm) and microve-\nsicles (100 – 1000 nm) which are either actively released\nfrom cells by fusion of multivesicular bodies to plasma\nmembrane or formed by direct budding of the cell mem-\nbrane into the extracellular space, respectively. Exosomes\nand microvesicles contain proteins, lipids and nucleic\nacids such as mRNA and miRNA from their cell of origin.\nThe increasing evidence of their roles in cell-to-cell\ncommunication [ 8, 9], their high abundance in plasma\n(1012 per mL), and highly stable nature, are several of the\nreasons for the increased interest to identify EV-based bio-\nmarkers. In ovarian cancer, EVs have been explored in as-\ncites and urine for miRNA and protein surface markers\n[10, 11]. Because of the demonstrated roles of EVs in com-\nmunication and tumor progression, we initiated a study to\nexamine the expression of EV mRNA and miRNA in ovar-\nian cancer ascites and compared the expression with an\nideal but difficult to obtain control source, peritoneal\nfluids from females inflicted with benign gynecologic dis-\neases. These studies indicate that malignant ascites EVs\npackage quantifiable mRNA and miRNA that can poten-\ntially provide insights into diagnostic biomarkers and\ntherapeutic targets.\nMethods\nStudy design\nThe experimental design is separated into two parts: 1)\npreliminary characterization of clinical samples and 2)\nevaluations of the biomarkers in cultured cell line. For\nthe characterization of clinical samples, four main studies\nwere performed: 1) size and concentration of EVs, 2) pre-\nliminary qPCR screening of mRNA biomarkers identified\nthrough previous literature, 3) pilot RNA-sequencing of\nEV mRNA, and 4) RT-qPCR validation of mRNA and\nscreening of miRNA. The mRNA biomarkers identified\nthrough the RNA-seq qPCR validation were evaluated in\nseveral appropriate cell lines.\nBiofluid collection\nAscites from advanced stage ovarian cancer patients\nwere collected from the peritoneal cavity using Yankauer\nsuction connected to a drainage bag, or bulb suction\nfrom volumes smaller than 500 mL. Bulb suction was\nalso used to collect small volumes (1 – 5 mL) of\nperitoneal fluids from patients with non-malignant con-\nditions. Samples were collected at the Brigham and\nWomen’ s hospital under informed consent and Internal\nReview Board approval. Fluids were transferred into\n50 cm 3 tubes and centrifuged at 2000 x g for 15 min at\n4 °C to remove cell debris. The supernatant was then\nstored at − 80 °C until further use.\nCell culture\nImmortalized human fallopian tube secretory epithelial cell\nline (FTSEC) was kindly provided by Dr. Ronny Drapkin\n[12]. Ovarian surface epithelial (OSE7, HOSE1 – 15) and\nhigh-grade serous ovarian cancer (SKOV3, OVCA3) cell\nlines have been previously described [ 13]. FTSEC cells\nwere cultured in DMEM/Ham ’ s F-12 1:1 (Cellgro,\nMediatech, Inc. Manassas, VA) supplemented with 2%\nUltroser G serum substitute (Pall Corp., Port Washington,\nNY). Ovarian epithelial cell lines were cultured in a mix-\nture of medium 199 and MCDB105 medium (1:1) (Sigma,\nSt. Louis, MO) supplemented with 10% fetal bovine serum\n(FBS, Invitrogen, Carlsbad, CA).\nStandard media was replaced with exosome-free fetal\nbovine serum (System Biosciences, SBI, Palo Alto, CA)\ncontaining media 24 h prior to conditioned media col-\nlection. Cell counts were determined at the time of con-\nditioned media collection and ranged from 2 to 9 × 10 5\ncells/mL. Conditioned media were transferred to 50 cm 3\ntubes and centrifuged at 2000×g for 15 min at 4 °C to\nremove cells and cell debris. The supernatant was then\nstored at − 80 °C until further use.\nDifferential ultracentrifugation and EV characterization\nAscites and peritoneal fluid samples were initially centri-\nfuged at 2000 x g for 10 min to remove large debris. The\nsupernatant was collected and further centrifuged at\n10,000 x g for 30 min. EVs in the supernatant were then\ncollected by ultracentrifugation at 100,000 x g for 1 h,\nwashed with PBS, then collected again at 100,000 x g for\n1 h in a Ti90 rotor. EV pellet was resuspended in PBS\nand stored at − 80 °C until further use. Nanoparticle\ntracking analysis of the EVs was conducted by Nanosight\nLM10 (Particle Characterization Laboratories, Inc.,\nNovato, CA). Samples were diluted from 1:5 to 1:50 and\napplied to ELISA assay for EpCAM detection (Thermo\nScientific, Frederick, MD) at 450 nm.\nScanning electron microscopy (SEM)\nAscites samples were pre-centrifuged at 3000 x g for\n15 min at 4 °C before applying to the ExoComplete fil-\nterplate (Hitachi Chemical Diagnostics, Inc., (HCD),\nMountain View, CA). EVs were captured on the filter\nmembrane after centrifugation at 2000 x g for 5 min.\nFilters were fixed with 4% paraformaldehyde, and\nblocked with casein before incubation with 1:200 anti-\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 2 of 9\n\nhuman CD63 (Clone H5C6, BioLegend, Dedham, MA)\nfor 1 h with gentle shaking. Samples were washed 3\ntimes with casein PBS followed by incubation with 1:40\ngoat anti-mouse IgG gold colloid 9.0 – 11.0 nm (Sigma-\nAldrich, St. Louis, MO) for 2 h. EVs were washed 3\ntimes with casein PBS and PBS. Samples were fixed\nagain with 4% paraformaldehyde for 5 min. Samples\nwere washed once with PBS and 2 times with distilled\nwater before incubation with 100 μL Silver Enhancement\nsolution (BBI Solutions) for 10 min. Filters were washed\nwith distilled water and air-dried overnight before\nanalysis with SEM using Hitachi S-4800.\nNext generation RNA sequencing\nAscites ( n = 2) and peritoneal fluid ( n = 3) samples were\ncentrifuged at 2000 x g for 10 min at 4 °C after thawing as\ndescribed above. The ExoComplete filterplate (HCD) cap-\ntured EVs from 400 μL of centrifuged ascites and periton-\neal fluid supernatant. Total RNA from EVs attached to the\nfilter membrane was isolated using miRNeasy kit (Qiagen,\nValencia, CA). Total RNA was monitored for quality con-\ntrol using the Agilent Bioanalyzer Nano RNA chip and\nNanodrop absorbance ratios for 260/280 nm and 260/\n230 nm. Library construction was performed according to\nthe Illumina TruSeq mRNA stranded protocol. The input\nquantity for total RNA was within the recommended\nrange and mRNAs and noncoding RNAs with poly(A)\ntails was enriched using oligo dT magnetic beads. The\nenriched poly(A)+ RNA was chemically fragmented. First\nstrand synthesis used random primers and reverse tran-\nscriptase to make cDNA. After second strand synthesis,\nthe ds cDNA was cleaned using AMPure XP beads, cDNA\nwas end repaired and then the 3 ′ ends were adenylated.\nIllumina barcoded adapters were ligated on the ends and\nthe adapter ligated fragments were enriched by nine cycles\nof PCR. The resulting libraries were validated by qPCR\nand sized by Agilent Bioanalyzer DNA high sensitivity\nchip. Concentrations for the libraries were normalized and\nthen multiplexed together. Multiplexed libraries were se-\nquenced using paired end 100 cycles chemistry for the\nHiSeq 2500. The version of HiSeq control software was\nHCS 2.2.58 with real time analysis software, RTA 1.18.64.\nFASTQ files were input into Maverix Biomics platform\nmRNA-seq for differential expression in eukaryotes ver-\nsion 2.5. Additionally, Ingenuity Pathway Analysis (IPA)\n(Qiagen) was employed to identify biological pathway\nmodulation.\nExtracellular vesicle mRNA analysis\nAscites ( n = 8), peritoneal fluid ( n = 10), and cell culture\nconditioned media (2 mL) were processed by first thawing\nfor 10 min at 37 °C and then placed on ice. For preliminary\nqPCR screening of mRNA in clinical samples, ascites ( n =\n8) and peritoneal fluid ( n = 2) were used. Thawed samples\nwere centrifuged at 2000 x g for 10 min at 4 °C. Three hun-\ndred fifty μL of supernatant was applied to ExoComplete\nfilterplate (HCD) and centrifuged. For cell culture condi-\ntioned media, samples were centrifuged as above and su-\npernatants were applied to EV collection tubes (HCD). EVs\nwere then captured onto the filter membrane after repeated\ncentrifugation. From this step, procedures for mRNA ana-\nlysis for ascites, peritoneal fluid, and cell culture condi-\ntioned media are identical. Exocomplete lysis buffer is\napplied to the EVs captured on the filter and incubated at\n37 °C for 10 min. Centrifugation of the filterplate or filter\ntips from the collection tube was performed at 2000 x g for\n5m i na t4° Ct ot r a n s f e rl y s a t et ot h em R N AC a p t u r eP l a t e\nfor hybridization of mRNA to the oligo(dT)-covalently\nlinked wells. After wash steps, on-plate random-primed\ncDNA synthesis using MMLV was performed at 37 °C for\n2 h. For qPCR analysis, 2 μL of cDNA was used with Sso\nAdvanced SYBR mix (Bio-Rad, Hercules, CA) and gene-\nspecific primers (Additional file1). Real-time PCR was per-\nformed on a ViiA7 (Thermo Fisher Scientific, Inc., Wal-\ntham, MA) instrument using the following profile: initial\ndenaturation at 95 °C for 10 min, 40 cycles of 95 °C for\n30 s and 65 °C for 1 min, melting curve analysis. Ct values\ngreater than 36 were set to 36 cycles for data analysis.\nReal-time PCR data was processed by Data Assist v3.01\n(Thermo Fisher Scientific, Inc.) and analyzed by Excel.\nExtracellular vesicle miRNA analysis\nAscites ( n = 8) and peritoneal fluid ( n = 10) were thawed\nfor 10 min at 37 °C and then placed on ice. Four hun-\ndred μL ascites and peritoneal fluid were centrifuged at\n2000 x g for 10 min at 4 °C. Supernatant was applied to\nExoComplete filterplate and centrifuged at 2000 x g for\n5 min at 4 °C. Lysis buffer from miRNeasy (Qiagen), was\napplied to the wells in the filterplate. Total RNA was\nisolated per manufacturer ’ s protocol. Synthesis of cDNA\nwas performed using miScript RT kit (Qiagen). The\ncDNA was diluted 1:4 and 1 μL was used with the\nmiScript PCR assay (Qiagen) for qPCR screening. For\nmiRNA analysis, Excel and DataAssist v3.01 was used\nwith the following parameters: cut-off value of Ct = 36,\nSNORD61 selected as endogenous reference RNA.\nStatistical analysis\nThe statistical analysis was performed using Microsoft\nExcel software. The statistical significance of the differ-\nences was determined by applying the Student ’ s t-test.\nResults\nPatient characteristics and samples\nThe patient characteristics and clinical information were\nobtained for ovarian cancer ascites samples\n(Additional file 2). All samples were collected at the time\nof diagnosis. Eight patients were diagnosed with serous\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 3 of 9\n\ntype ovarian cancer: seven with high grade, and a single\npatient was diagnosed with low grade serous type. The\nages ranged from 48 to 80 years with an average age at\ndiagnosis of 64 ± 12 years old. Using the available data,\nCA125 levels were elevated with average values at 1289\n± 541 U/mL at diagnosis and average progression-free\nand overall survival were 19 ± 8 and 33 ± 22 months, re-\nspectively. For peritoneal fluid samples, no malignant\ncells were identified in peritoneal washings or from\novary, fallopian tube, uterus and cervix pathology report\n(Additional file 3).\nAscites and peritoneal fluid EV characterization\nAscites and peritoneal fluid extracellular vesicles were\nisolated using differential ultracentrifugation and charac-\nterized for size and concentration using nanoparticle\ntracking analysis (Table 1). This preliminary\ncharacterization indicated average sizes of ascites and\nperitoneal fluid EVs were not statistically different and\naveraged 113 ± 11.5 nm. CD63-positive particles <\n200 nm in diameter were observed in scanning electron\nmicrographs of ovarian cancer ascites samples applied to\nthe ExoComplete EV capture filterplate and correlate to\nsize estimates from nanoparticle tracking analysis (Fig. 1).\nEV concentrations ranged from 10 10 to 10 12particles/mL\nand were also not statistically significant between the\ntwo sample types. EVs from both ascites and peritoneal\nfluids were also evaluated for EpCAM surface marker\nexpression by ELISA. EpCAM is a proposed surface\nmarker of ovarian cancer-derived exosomes. Only one of\neach sample type was found to be positive for EpCAM.\nThe single ascites and peritoneal fluid-derived EV sam-\nples had an EpCAM concentration of 670 pg/mL and\n1.44 ng/mL, respectively. The other samples were below\nthe limit of detection (50 pg/mL) for the ELISA assay.\nNext generation RNA sequencing and qPCR validation\nA literature search for potential ovarian cancer bio-\nmarkers identified 50 mRNA candidates (Additional file 4)\nwhich were then evaluated in a preliminary qPCR\nscreening of ovarian cancer ascites ( n = 8) and peritoneal\nfluid EVs ( n = 2). Three mRNAs, NANOG, SPINT2,\nZEB2, were found to be significantly elevated (> 2-fold,\np-value < 0.05) in ascites in the preliminary screening\n(Additional file 5). To further identify differentially\nexpressed genes and additional mRNA markers, next\ngeneration poly(A) + RNA sequencing was performed on\nselected samples of ascites ( n = 2) and peritoneal fluid\n(n = 3). RNA sequencing mapping percentages to hg19\nassembly ranged from 50% to 87%, and greater than 70%\nof the reads were kept after quality assessment\n(Additional file 6). The read distributions indicate that\nboth samples have a higher percentage of reads mapping\nto exons and 3 ’ UTR compared to introns and 5 ’ UTR\nas expected based on sequencing preparation method-\nology (Additional file 7). Intergenic regions ranged from\n2.5– 15.9% for peritoneal fluid EVs and 24 – 47% for\novarian cancer ascites EVs.\nPathway analysis of RNA-seq differential gene expres-\nsion data from ovarian cancer ascites and peritoneal fluid\nEVs identified organismal injury, cancer and reproductive\nsystem diseases as the main categories of diseases and dis-\norders (Additional file 8). In terms of function, results\nwere consistent with the biological environment of an ad-\nvanced stage disease. Signaling pathways within growth,\nmalignant tumor and advanced malignancy categories\nwere predicted to be up-regulated whereas tumor cell\ndeath pathways were predicted to decline.\nRNA-seq differential gene expression analysis identi-\nfied 114 genes with statistical significance ( p < 0.05)\n(Additional file 9). From this list, 30 genes selected for\nqPCR validation based on fold change, p-value, abun-\ndance and function were measured in eight ovarian can-\ncer ascites and ten peritoneal fluid samples\n(Additional file 10). SPINT2, one of the three mRNA\nfound to be differentially expressed in the initial screen\nbased on literature-identified biomarkers was identified\nin the RNA-seq analysis. The remaining two genes,\nNANOG and ZEB2, were added to the list of 30 genes in\nthe final RNA-seq qPCR validation assays . ACTB was se-\nlected as a reference gene for mRNA normalization. Of\nthe selected mRNA for qPCR validation, five were found\nto be significantly ( p < 0.05) differentially expressed in\novarian cancer ascites and peritoneal fluid (Fig. 2). Three\nTable 1 Ascites ( n = 3) and peritoneal fluid ( n = 3) EV characteristics\nEV Sample Source a ID EpCAM ELISAa Concentration (particles/mL) Size (nm)\nAscites A2 positive 2.17E + 10 105\nAscites A3 negative 3.20E + 12 106\nAscites A10 negative 5.43E + 10 128\nPeritoneal Fluid B7 negative 1.20E + 10 106\nPeritoneal Fluid B11 positive 8.47E + 11 106\nPeritoneal Fluid B13 negative 1.55E + 10 128\naEV isolated by differential centrifugation\nb10E9 – 10E11 EVs applied\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 4 of 9\n\nmRNAs, CA11, LAMA4, MEDAG , were .01 – .28-fold\nlower expressed and two mRNA, SPINT2 and NANOG,\nwere 3.2 – 5.8-fold higher expressed in ovarian cancer\nascites versus peritoneal fluid EVs.\nAscites and peritoneal fluid EV miRNA analysis\nSmall RNAs are abundant within extracellular vesicles.\nHere, we quantitate specific miRNA previously identified\nto be involved in ovarian cancer progression or invasive-\nness [ 14– 18]. There were six miRNAs, let-7b, miR23b,\nmiR29a, miR30d, miR205, miR720 , which were found to\nbe significantly (p < 0.05) decreased .01 – .21-fold in ovar-\nian cancer ascites compared to benign peritoneal fluid\nwhen normalized to SNORD61 (Fig. 3).\nMultivariate discriminate analysis\nThe combined mRNA and miRNA raw qPCR data were\nused in multivariate discriminate analysis (Additional file11).\nThe predictors , LAMA4, CA11, MEDAG, NANOG,\nSPINT2, let-7b, miR23b, and miR29a were found to be suf-\nficient to classify 87.5% and 100% of ovarian cancer ( n =8 )\nand disease control (n = 10) groups, respectively.\nSpecific EV mRNA from normal and cancer cell lines\nThe EVs released from normal human fallopian tube epi-\nthelial (FTSEC194), ovarian surface epithelial (OSE7,\nHOSE1– 15), and high grade serous ovarian cancer\n(SKOV3, OVCA3) cell lines were analyzed for specific\nmRNA that were differentially expressed in ovarian\nAB\nFig. 1 a,b Scanning electron microscopy of ovarian cancer ascites EVs captured on the membrane of the ExoComplete filterplate. Ovarian cancer\nascites was applied to the filterplate, fixed, labeled with anti-CD63 primary monoclonal antibody and anti-mouse IgG colloidal gold with silver\nenhancement. Black arrows indicate selected extracellular vesicles with diameters < 200 nm captured on the membrane. Images were obtained\nwith accelerating voltages of 2.0 kV in detection mode\nFig. 2 Relative gene expression in peritoneal fluids (n = 10) and ovarian cancer ascites (n =8 )E Vs a m p l e s .a-e CA11, MEDAG, LAMA4, SPINT2, NANOG\nnormalized to ACTB are shown as 2^-ΔCT gene expression levels with average and SD indicated by horizontal lines. Maximum gene expression was\nset at a cut-off of 10. Gene expression values for each individual subject are represented as solid circle and square symbols for ascites and peritoneal\nfluids, respectively. Statistical significance for (a-d)i s p < 0.05 and (e)i s p <0 . 0 0 5u s i n gS t u d e n t’ st - t e s t\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 5 of 9\n\ncancer ascites and peritoneal fluids (Fig. 4). The immor-\ntalized normal fallopian and ovarian surface epithelial\ncells were used as controls. Using 3 mL of conditioned\nmedia from each cell line, we confirm that CA11 mRNA\nis 0.1 – 0.5-fold lower abundance and NANOG is 50-fold\nhigher abundance in EVs released from high-grade ser-\nous ovarian cancer cells, OVCA3 . Although CA11\nexpression appeared to be elevated in SKOV3 cells,\nmRNA levels for CA11 as well as NANOG were not sta-\ntistically significant compared to normal cells. SPINT2,\nhowever, was found to be significantly elevated in both\nSKOV3 and OVCA 3 cells compared to normal cells.\nLAMA4 was expressed in 0.1-fold lower abundance in\nOVCA3, but up to 2-fold higher abundance in EVs\nFig. 3 Relative miRNA expression levels in peritoneal fluids ( n = 10) and ovarian cancer ascites ( n = 8) EV samples. a-f Let7b, miR205, miR23b,\nmiR29a, miR30d, miR720 normalized to SNORD61 are shown as 2^- ΔCT expression levels with average and SD indicated by horizontal lines. Gene\nexpression values for each individual subject are represented as solid circle and square symbols for ascites and peritoneal fluids, respectively.\nStatistical significance for ( a-d)i s p < 0.05 and ( e)i s p < 0.005 using Student ’ s t-test\nFig. 4 Specific mRNA expression normalized to ACTB from immortalized fallopian (FTSEC194), ovarian surface epithelial (OSE7, HOSE1– 15) and ovarian\ncancer cell lines (SKOV3, OVCA3) EVs released in conditioned media. a SPINT2,( b) NANOG,( c) CA11,( d) LAMA4 mRNA quantitation (2^-ΔCT) is shown\nas column graph with average and SD ( n = 2) and statistical significance indicated by a bar representing p < 0.05 by Student’ s t-test\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 6 of 9\n\noriginating from SKOV3 cells compared to ovarian sur-\nface epithelial cell-derived EVs (HOSE1 – 15). The two\nhigh grade serous ovarian cancer cell lines, SKOV3 and\nOVCA3, demonstrated similar expression patterns for\nSPINT2 and NANOG, but distinct relative expression\nfor CA11 and LAMA4 . MEDAG mRNA was present\nin very low levels and was not reproducibly detected\nin all cell lines.\nDiscussion\nExtracellular vesicles (EVs), including exosomes and\nmicrovesicles, are small membranous particles released\nfrom all cells and found in many biofluids. These vesi-\ncles are thought to provide a mode of cellular communi-\ncation and deliver their cargo of protein, DNA and\nRNA, to target cells [ 10, 11]. As a result, EVs have been\na novel source of biomarkers in a wide range of diseases.\nIn this study, we analyzed EV mRNA and miRNA from\novarian cancer ascites and benign peritoneal fluids to de-\ntermine if they can provide biological insight into metas-\ntasis and be a potential source of novel diagnostic\nbiomarkers.\nThe average size and concentrations of EVs isolated\nfrom ovarian cancer ascites and peritoneal fluids were <\n120 nm and at least 10 10 particles/mL and are within\nrange of EVs isolated from other biofluids such as\nplasma and urine. The variability in particle concentra-\ntion observed between samples is consistent with what\nhas been observed in previous literature [ 11]. In\naddition, EpCAM has been proposed as a marker for\novarian cancer-derived exosomes [ 11], but only 1 out of\n3 of the ovarian cancer and peritoneal fluid EV samples\nwere positive for EpCAM. This may be reflective of the\ntypically low presence of adenocarcinoma cells (< 0.1%)\nwithin ovarian cancer ascites and the presence of endo-\nmetrial epithelial cells in peritoneal fluids [ 19, 20]. The\novarian cancer ascites EV sample did not demonstrate a\nhigher EpCAM concentration compared to the periton-\neal fluid EV sample.\nThe ovarian cancer ascites and peritoneal fluid EVs\nwere analyzed for differential mRNA expression. The\nmRNAs in this study were selected based on a combin-\nation of next generation sequencing (NGS) results and\nprevious literature. The EV characterization and RNA\nsequencing were pilot analyses of the EVs and identifica-\ntion of new targets. IPA results confirmed RNA-seq is a\nmethodology which may identify relevant biomarkers for\ndiagnosis. The qPCR validation of the RNA-seq results\nemployed more samples to confirm the gene expression\npattern. Based on these qPCR analyses, the mRNAs con-\nfirmed to be decreased in ovarian cancer ascites com-\npared to peritoneal fluid EVs included LAMA4, CA11 ,\nand MEDAG. EVs from a high grade serous ovarian\nadenocarcinoma cell line OVCA3 also contained less\nabundance of CA11 compared to controls from immor-\ntalized epithelial fallopian and ovarian cells. Recently,\nfallopian tube secretory epithelial cells have been pro-\nposed to be the precursor tissue for high grade serous\novarian cancer, and immortalized human fallopian tube\nsecretory epithelial cells are being used for studying\nearly-stage development of high grade serous ovarian\ncancer [ 21]. Based on the publicly available genotype-\ntissue expression (GTEx) database, CA11 mRNA has\nhigh baseline expression in brain and medium expres-\nsion in tissues such as ovary. Although no previous stud-\nies were found to link CA11 with gynecological cancers,\nthis mRNA was down-regulated in human gastric cancer\n[22]. CA11 is a member of carbonic anhydrase family\nknown to participate in biological processes such as for-\nmation of aqueous humor, CSF and saliva. Similar to\nCA11, MEDAG was in lower abundance in ovarian can-\ncer ascites EVs. MEDAG expression, however, was either\nat the limit of detection or undetectable in the fallopian\nand ovarian cell lines used in this study suggesting low\nbaseline expression of MEDAG in these tissues. The\nGTEx database confirms this observation and shows\nMEDAG expressed at higher levels in visceral adipose\nand arterial tissues. MEDAG is a gene involved in pro-\ncesses that promote adipocyte differentiation, lipid accu-\nmulation, and glucose uptake in mature adipocytes.\nBecause ovarian cancer cells bind preferentially to\nomental fat and use human omental adipocytes as an en-\nergy source, the decrease observed in ovarian cancer as-\ncites MEDAG EV mRNA may reflect changes in the\nascites microenvironment [ 23]. A previous study using\nmicroarray has also shown a lower MEDAG and LAMA4\nexpression in ovarian cancer compared to normal epithe-\nlial cells [ 24]. LAMA4 is a member of the major family of\nnon-collagenous constituents of basement membranes.\nThe decrease in LAMA4 observed in malignant ascites EV\nand OVCA3 EV , in contrast to the increase in SKOV3 EV ,\nare interesting and warrants further experimental studies\nto evaluate the relationship. Ascites EVs could be\nproviding cell-to-cell communication through unique\nmolecular profiles promoting cancer cell survival within\nthe ascites milieu.\nIn contrast to LAMA4, CA11 ,a n d MEDAG,t h e2\nmRNAs, SPINT2 and NANOG, were found to be increased\nin ovarian cancer ascites compared to peritoneal fluid EVs.\nSPINT2 was also in higher abundance in OVCA3 and\nSKOV3 EVs compared to controls. Interestingly, although\nSPINT2 is a putative suppressor, we demonstrate an in-\ncrease in cancer ascites [ 25]. Müller-Pillasch et al. also re-\nports that SPINT2 expression was elevated in pancreatic\ncancer [26]. NANOG, on the other hand, is a DNA bind-\ning homeobox transcription factor involved in embryonic\nstem cell proliferation, renewal and pluripotency. NANOG\nis increased in expression from normal tissue, benign,\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 7 of 9\n\nborderline, and malignant tumors of ovarian serous cysta-\ndenocarcinomas and protein is found selectively associated\nwith high-grade ovarian serous carcinoma [ 27, 28].\nNANOG mRNA was found to be significantly increased in\nOVCA3 EVs compared to OSE7 control. Based on previ-\nous studies relating to these mRNA, CA11 and MEDAG\nmay be involved in maintaining malignant ascites micro-\nenvironment, while LAMA4, NANOG and SPINT2 activ-\nities could regulate ovarian cancer progression and\nmetastasis.\nThere were six miRNA biomarkers , let7b, miR205,\nmiR23b, miR29a, miR30d , and miR720, significantly\ndown-regulated in ovarian cancer ascites EVs compared\nto peritoneal fluids. These miRNAs have previously been\nshown to be involved in ovarian cancer progression, in-\nvasion or metastasis. The let-7 family regulates down-\nstream gene targets involved in self-renewal of\nmesenchymal stem cells derived from human embryonic\nstem cells . Let-7b is often dysregulated in ovarian cancer\nand is associated with poor prognosis [ 29]. Recently,\nmiR205 was shown to be elevated in ovarian cancer tis-\nsue and associated with tumor growth and metastasis in\novarian cancer [ 30]. In pre-surgical plasma , miR720 was\nelevated in women who had short overall survival (<\n2 years) when compared to women with long overall\nsurvival (> 4 years) after their diagnosis [ 31]. In contrast,\nmiR23b expression is lower in epithelial ovarian carcin-\noma and borderline tumors than in normal ovarian tis-\nsues and benign tumors consistent with the lower\nabundance observed in ovarian cancer ascites in this\nstudy. MiR23b was shown to target cyclin G1 and sup-\npress ovarian cancer tumorigenesis and progression [ 14].\nMiR29a was also shown to have tumor suppressive ef-\nfects and may contribute to cisplatin resistance of ovar-\nian cancer cells [ 32, 33]. MiR30d also functions as a\nsuppressor of ovarian cancer progression notably by de-\ncreasing Snail expression and blocking TGF-b1-induced\nEMT process [ 34].\nMalignant ascites presents in approximately 30% of\nwomen with ovarian cancer and is frequently tapped to\nrelieve symptoms. This fluid is composed of lympho-\ncytes, epithelial cells, and EVs, and provides clues into\nascites formation and metastatic progression. Functional\nanalysis of each specific mRNA and miRNA will be\nneeded to determine the biological significance of their\ndifferential expression in ovarian cancer.\nConclusions\nHere, we demonstrate that EVs from ovarian cancer as-\ncites contain distinct RNA expression signatures from\nbenign peritoneal fluids, control samples that are rarely\navailable to research. The two mRNA markers SPINT2\nand NANOG that are upregulated in cancer ascites\nrelative to peritoneal fluids may have potential as diag-\nnostic biomarkers. Through continued liquid biopsy in-\nvestigations, an understanding of the mechanisms\ninvolved in advanced stage disease and development of\nchemo-resistant disease may lead to alternative thera-\npeutic targets and improved palliative care.\nAdditional files\nAdditional File 1: Primer sequences (5 ′ to 3 ′) for qPCR validation.\n(DOCX 13 kb)\nAdditional File 2: Clinical information from patient ascites samples.\n(DOCX 13 kb)\nAdditional File 3: Clinical information from benign peritoneal fluid (PF)\nsample pathology reports. (DOCX 12 kb)\nAdditional File 4: Preliminary set of genes with primer sequences for\nscreening based on literature search. Additional genes ( MET, EGFR,\nEPCAM, CLDN3 ) were quantitated using Qiagen Quantitect Primer\nAssays. (DOCX 13 kb)\nAdditional File 5: Volcano plot displays p-values vs. fold change of\novarian cancer ascites ( n = 8) and peritoneal ( n = 2) EVs. Three mRNA\n(NANOG, SPINT2, ZEB2 ) show values above the fold change boundary of 2\n(2-fold change) and a p-value of 0.05. Plot generated using. Data Assist\nv3.01 software. (PPTX 58 kb)\nAdditional File 6: mRNA sequencing reads and percentage aligned to\nhg19 assembly. (DOCX 12 kb)\nAdditional File 7: The distribution of EV mRNA sequencing reads\nmapping to human genome annotations. The percentage of reads (ave.\n± SD) overlapping genomic features including exons, introns, UTR, and\nintergenic regions are shown for peritoneal fluids ( n = 3) and ovarian\ncancer ascites samples ( n = 2). (PPTX 115 kb)\nAdditional File 8: Ingenuity Pathway Analysis summary of top diseases\nand disorders, top canonical pathways, and top molecular and cellular\nfunctions. P-values are calculated using the right-tailed Fisher Exact Test\nand number of molecules are based on Ingenuity Knowledge Base with\ninformation contained in Canonical Pathways coming from specific\njournal articles, review articles, textbooks and HumanCyc. (DOCX 22 kb)\nAdditional File 9: Differentially expressed genes from RNA sequencing\nanalysis. Significantly increased RNA ( p < 0.05) from ovarian cancer ascites\n(n = 2) compared to benign peritoneal fluid (n = 3) are listed below as\neither up-regulated or down-regulated in ascites compared to peritoneal\nfluids. (DOCX 22 kb)\nAdditional File10: Next generation RNA sequencing data were plotted\nfor each sample of ovarian cancer ascites (OC) and benign peritoneal fluids\n(DC). RPKM values for each corresponding gene are indicated in blue and\nthe 30 selected genes for qPCR validation are labeled in red. Solid blue lines\nindicate linear regression. Both over- and under-expressed genes were\nselected for validation. (PPTX 318 kb)\nAdditional File 11: Multivariate discriminant analysis of mRNA and\nmiRNA qPCR data. Linear discriminant functions are listed for ovarian\ncancer and disease control groups. (DOCX 12 kb)\nAbbreviations\nCA125: Cancer antigen 125; EOC: Epithelial ovarian cancer; EV: Extracellular\nvesicles; FTSEC: Fallopian tube secretory epithelial cell\nAcknowledgements\nN/A\nFunding\nNo external source of funding was used for this study.\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 8 of 9\n\nAvailability of data and materials\nThe datasets used and/or analysed during the current study are available\nfrom the corresponding author on reasonable request.\nAuthors’ contributions\nCMY and SWN designed the study and contributed to writing the\nmanuscript. CMY, MO, and TM collected, analyzed and/or interpreted data.\nMM and RB collected peritoneal and ascites samples. All authors read and\napproved final manuscript.\nEthics approval and consent to participate\nHuman samples were collected at Brigham and Women ’ s hospital under\ninformed consent and Internal Review Board approval.\nConsent for publication\nN/A\nCompeting interests\nCMY, MO, and TM are employees of Hitachi Chemical Co. America, Ltd. R&D\nCenter and the two authors, CMY and SWN have submitted a provisional\npatent (Appl. No. 62/507,091) entitled, “Methods for detecting ovarian cancer\nusing extracellular vesicles for molecular analysis ”.\nPublisher’sN o t e\nSpringer Nature remains neutral with regard to jurisdictional claims in\npublished maps and institutional affiliations.\nAuthor details\n1Hitachi Chemical Co. America, Ltd. R and D Center, 1003 Health Sciences Rd,\nIrvine, CA 92617, USA. 2Department of Obstetrics, Gynecology and\nReproductive Biology, Gynecologic Oncology Division, Brigham and\nWomen’ s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA\n02115, USA. 3Department of Obstetrics and Gynecology, Tuft Medical Center,\n800 Washington Street, Boston, MA 02111, USA.\nReceived: 2 November 2017 Accepted: 22 February 2018\nReferences\n1. U.S. Cancer Statistics Working Group. United States cancer statistics:\n1999– 2013 incidence and mortality web-based report. https://www.cdc.\ngov/cancer/ovarian/statistics/index.htm (accessed 17 Aug 2017).\n2. National Cancer Institute: Surveillance, Epidemiology, and End Results Program.\nhttps://seer.cancer.gov/statfacts/html/ovary.html(accessed 17 Aug 2017).\n3. Skates SJ, Xu FJ, Yu YH, Sjovall K, Einhorn N, Chang Y, et al. Toward an\noptimal algorithm for ovarian cancer screening with longitudinal tumor\nmarkers. Cancer. 1995;76(Suppl 10):2004 – 10.\n4. Streppel MM, Vincent A, Mukherjee R, Campbell NR, Chen SH,\nKonstantopoulos K, et al. Mucin 16 (cancer antigen 125 ) expression in\nhuman tissues and cell lines and correlation with clinical outcome in\nadenocarcinomas of the pancreas, esophagus, stomach, and colon. Hum\nPathol. 2012;43:10.\n5. Jelovac D, Armstrong DK. Recent progress in the diagnosis and treatment\nof ovarian cancer. CA Cancer J Clin. 2011;61:3.\n6. Buys SS, Partridge E, Black A, et al. Effect of screening on ovarian cancer\nmortality: the prostate, lung, colorectal and ovarian (PLCO) cancer screening\nrandomized controlled trial. JAMA. 2011;305:22.\n7. Dorayappan KDP, Wallbillich JJ, Cohn DE, Selvendiran K. The biological\nsignificance and clinical applications of exosomes in ovarian cancer.\nGynecol Oncol. 2016;142:1.\n8. Crow J, Atay S, Banskota S, Artale B, Schmitt S, Godwin AK. Exosomes as\nmediators of platinum resistance in ovarian cancer, Oncotarget. 201; doi:\nhttps://doi.org/10.18632/oncotarget.14440.\n9. Nakamura K, Sawada K, Kinose Y, Yoshimura A, Toda A, Nakatsuka E, et al.\nExosomes promote ovarian cancer cell invasion through transfer of CD44 to\nperitoneal mesothelial cells. Mol Cancer Res. 2017;15:1.\n10. Zhou J, Gong G, Tan H, Dai F, Zhu X, Chen Y, et al. Urinary microRNA-30a-5p is\na potential biomarker for ovarian serous adenocarcinoma. Onc Rep. 2015;33:6.\n11. Im H, Shao H, Park YI, Peterson VM, Castro CM, Weissleder R, et al. Label-free\ndetection and molecular profiling of exosomes with a nano-plasmonic\nsensor. Nat Biotechnol. 2014;32:5.\n12. Karst AM, Levanon K, Drapkin R. Modeling high-grade serous ovarian\ncarcinogenesis from the fallopian tube. PNAS. 2011;108:18.\n13. Zhang Y, Kwok JS-L, Choi P-W, Liu M, Yang J, Singh M, et al. Pinin interacts\nwith C-terminal binding proteins for RNA alternative splicing and epithelial\ncell identity of human ovarian cancer cells. Oncotarget. 2016;7:10.\n14. Yan J, Jiang J-Y, Meng X-N, Xiu Y-L, Zong Z-H. MiR-23b targets cyclin G1\nand suppresses ovarian cancer tumorigenesis and progression. J Exp Clin Ca\nRes. 2016;35:31.\n15. Li J, Hu K, Gong G, Zhu D, Wang Y, Liu H, et al. Upregulation of miR-205\ntranscriptionally suppresses SMAD4 and PTEN and contributes to human\novarian cancer progression. Sci Reports. 2016;1:7.\n16. Meng X, Muller V, Milde-Langosch K, Trillsch F, Pantel K, Schwrzenbach H.\nDiagnostic and prognostic relevance of circulating exosomal miR-373,\nmiR-200a, miR-200b and miR-200c in patients with epithelial ovarian cancer.\nOncotarget. 2016;7:13.\n17. Yeung CLA, Co N-N, Tsuruga T, Yeung T-L, Kwan S-Y, Leung CS, et al.\nExosomal transfer of stroma-derived miR21 confers paclitaxel resistance in\novarian cancer cells through targeting APAF1. Nat Commun. 2016;29:7.\n18. Wu RL, Ali S, Bandyopadhyay S, Alosh B, Hayek K, Daaboul MHD, et al. Comparative\nanalysis of differentially expressed miRNAs and their downstream mRNAs in ovarian\ncancer and its associated endometriosis. J cancer Sci Ther. 2105;7:7.\n19. Kipps E, DSP T, Kaye SB. Meeting the challenge of ascites in ovarian cancer:\nnew avenues for therapy and research. Nat Rev Ca. 2103;13:4.\n20. O DF, Roskams T, Van den Eynde K, Vanhie A, Peterse DP, et al. The\npresence of endometrial cells in peritoneal fluid of women with and\nwithout endometriosis. Reprod Sci. 2017;24:2.\n21. Perets R, Wyant GA, Muto KW, Bijron JG, Poole BB, et al. Transformation of\nthe fallopian tube secretory epithelium leads to high-grade serous ovarian\ncancer in Brca;Tp53;Pten models. Cancer Cell. 2013;24:6.\n22. Yoshikawa Y, Mukai H, Hino F, Asada K, Kato I. Isolation of two novel genes,\ndown-regulated in gastric cancer. Jpn J Cancer Res. 2000;91:5.\n23. Nieman KM, Kenny HA, Penicka CV, Ladanyi A, Buell-Gutbrod R, Zillhardt MR,\net al. Adipocytes promote ovarian cancer metastasis and provide energy for\nrapid tumor growth. Nat Med. 2011;17:11.\n24. Yeung TL, Leung CS, Wong KK, Samimi G, Thompson MS, Liu J, et al. TGF-b\nmodulates ovarian cancer invasion by upregulating CAF-derived versican in\nthe tumor microenvironment. Ca Research. 2013;73:16.\n25. Kongkham PN, Northcott PA, Ra YS, Nakahara Y, Mainprize TG, Croul SE, et\nal. An epigenetic genome-wide screen identifies SPINT2 as a novel tumor\nsuppressor gene in pediatric medulloblastoma. Cancer Res. 2008; https://\ndoi.org/10.1158/0008-5472.CAN-08-2169.\n26. Muller-Pillasch F, Wallrapp C, Bartels K, Varga G, Friess H, Buchler M, et al.\nCloning of a new Kunitz-type protease inhibitor with a putative\ntransmembrane domain overexpressed in pancreatic cancer. Biochim\nBiophys Acta. 1998;1395:1.\n27. Pan Y, Jiao J, Zhou C, Cheng Q, Hu Y, Chen H. Nanog is highly expressed in\novarian serous cystadenocarcinoma and correlated with clinical stage and\npathological grade. Pathobiology. 2011;77:6.\n28. Kenda Suster N, Frkovic Grazio S, Virant-Klun I, Verdenik I, Smrkolj S. Cancer\nstem cell-related marker NANOG expression in ovarian serous tumors: a\nclinicopathological study of 159 cases. Int J Gynecol Cancer. 2017;9:2006.\n29. Zaman MS, Maher DM, Khan S, Jaggi M, Chauhan SC. Current status and\nimplications of microRNAs in ovarian cancer diagnosis and therapy. J\nOvarian Research. 2012;5:44.\n30. Li J, Hu K, Gong G, Zhu D, Wang Y, Liu H, et al. Upregulation of miR-205\ntranscriptionally suppresses SMAD4 and PTEN and contributes to human\novarian cancer progression. Sci Rep. 2017; https://doi.org/10.1038/srep41330.\n31. Shapira I, Oswald M, Lovecchio J, Khalili H, Menzin A, Whyte J, et al.\nCirculating biomarkers for detection of ovarian cancer and predicting\ncancer outcomes. Br J Cancer. 2014; https://doi.org/10.1038/bjc.2013.795.\n32. Yu PN, Yan MD, Lai HC, Huang RL, Chou YC, Lin WC, et al. Downregulation\nof miR-29 contributes to cisplatin resistance of ovarian cancer cells. Int J\nCancer. 2014; https://doi.org/10.1002/ijc.28399.\n33. Creighton CJ, Hernandez-Herrera A, Jacobsen A, Levine DA, Mankoo P,\nSchultz N, et al. Integrated analyses of microRNAs demonstrate their\nwidespread influence on gene expression in high-grade serous ovarian\ncarcinoma. PLoS One. 2012; https://doi.org/10.1371/journal.pone.0034546.\n34. Ye Z, Zhao L, Li J, Chen W, Li X. MiR-30d blocked transforming growth\nfactor β1-induced epithelial-mesenchymal transition by targeting snail in\novarian cancer cells. Int J Gyn Ca. 2015; https://doi.org/10.1097/IGC.\n0000000000000546.\nYamamoto et al. Journal of Ovarian Research  (2018) 11:20 Page 9 of 9","source_license":"CC0","license_restricted":false}