Clinical relevance and functional significance of cell-free microRNA-1260b expression profiles in infiltrative myxofibrosarcoma

Infiltrative tumor growth into adjacent soft tissues is a major cause of the frequent recurrence and tumor-related death of myxofibrosarcoma (MFS), but no useful biomarkers reflecting tumor burden and infiltrative growth are available. While emerging evidence suggests a diagnostic and functional role of extracellular/circulating microRNA (miRNA) in various malignant diseases, their significance in MFS patients remains unknown. Global miRNA profiling identified four upregulated miRNAs in MFS patient sera and culture media of MFS cells. Among these, serum miR-1260b level was significantly upregulated in patient serum discriminating from healthy individuals and closely correlated with clinical status and tumor dynamics in MFS-bearing mice. In addition, high miR-1260b expression in serum was correlated with radiological tail-like patterns, characteristic of the infiltrative MFS. The extracellular miR-1260b was embedded in tumor-derived extracellular vesicles (EVs) and promoted cellular invasion of MFS through the downregulation of PCDH9 in the adjacent normal fibroblasts. Collectively, circulating miR-1260b expression may represent a novel diagnostic target for tumor monitoring of this highly aggressive sarcoma. Moreover, EV-miR-1260b could act as a transfer messenger to adjacent cells and mediate the infiltrative growth of MFS, providing new insights into the mechanism of infiltrative nature via crosstalk between tumor cells and their microenvironment.


Results
Global microRNA profiling analysis of cfmiRNA in MFS patient serum and MFS cell culture media. To investigate the dysregulated expression of miRNAs in MFS patients' serum, miRNA microarray profiling analysis was performed using serum samples and culture media of MFS cells: serum collected from MFS patients (n = 5; Supplementary  Table 3), and MFS cell culture media (n = 1; NMFH-1). Among a total of 2006 analyzed miRNAs, 110 miRNAs were significantly upregulated in the serum collected from MFS patients (p < 0.05) compared to the serum collected from age-matched patients with benign tumors (NON), which included 89 miRNAs expressed in the MFS cell culture media (Supplementary Table 4). On the other hand, we identified 12 miRNAs that were significantly upregulated in the MFS serum samples (p < 0.05) compared to the serum collected from healthy individuals (CONT), in which 11 miRNAs had their expression in the MFS cell culture media (Supplementary Table 5). Among them, four upregulated miRNAs (miR-642a, miR-1260b, miR-4286, miR-4313) were commonly upregulated in the MFS serum samples, which all had their expression in the MFS cell culture media (Fig. 1A, Supplementary Table 6).
Upregulation of serum miR-1260b in MFS patients. We subsequently validated the results of microarray analysis by an independent qRT-PCR. Among four candidates, the expression levels of miR-642a and miR-4286 were significantly higher in serum collected from MFS patients than healthy individuals (p < 0.05; Fig. 1B). In addition, only miR-1260b expression was significantly upregulated in the serum of MFS patients compared with healthy individuals and benign tumor patients (p < 0.01; Fig. 1B). The expression levels of the miRNAs were also investigated in 10 pairs of tumor and normal tissue specimens. None of the four candidates was upregulated in MFS tissue specimens compared to the normal tissue specimens ( Supplementary Fig. 1).
We further analyzed the expression levels in the culture media of MFS cells and control hMSC cells. While miR-642a and miR-4286 expression levels were upregulated in the culture media of either NMFH-1 or NMFH-2, miR-1260b levels were significantly upregulated in the culture media of both NMFH-1 and NMFH-2 cells compared to hMSC (Fig. 1C). The extracellular miR-1260b expression levels in MFS cell culture media increased based on culture duration and cell number (Fig. 1D). The results indicated that miR-1260b was secreted into the extracellular spaces from the MFS cells. Therefore, we focused on this miRNA in further investigations.
Clinical relevance of serum miR-1260b levels in MFS patients. We further investigated the clinical relevance of serum miR-1260b levels to determine whether circulating miRNA could be used to monitor tumor dynamics in MFS patients. The serum samples collected at pre-and postoperative states from 10 MFS patients who underwent tumor wide resection were analyzed (Table 1). Eight patients were available for the administration of a pair of serum at the pre-and postoperative states. The serum miR-1260b levels significantly decreased postoperatively in all patients (p = 0.039; Fig. 2A). Subsequently, we compared the serum miR-1260b levels in patients with or without the tail-like pattern, which implies infiltrative growth of the disease [7][8][9] . Notably, serum miR-1260b expression levels at diagnosis in the sera from patients with the tail-like pattern (n = 6) were significantly higher than in those collected from patients without the tail-like pattern (solid pattern; n = 4; p = 0.038; Fig. 2B, Table 2). A receiver operating characteristic (ROC) curve was then generated to evaluate the sensitivity and specificity of serum miR-1260b as a biomarker for the infiltrative pattern in MFS patients. Circulating miR-1260b discriminated MFS with the tail-like pattern from MFS with the solid pattern with an area under the curve (AUC) value of 0.92 (95% confidence interval = 0.7281 to 1.1050; Fig. 2C). The sensitivity and specificity values for identifying MFS patients with the tail-like pattern were 83.3% and 75%, respectively. The results demonstrated that serum miR-1260b expression levels are associated with not only tumor burdens but also the infiltrative nature of MFS, indicating the usefulness of serum miR-1260b as a circulating biomarker for disease monitoring and the functional relevance of circulating miR-1260b in the infiltrative growth pattern of MFS.
Extracellular miR-1260b embedded in the tumor-derived EVs. In order to understand the intracellular and extracellular dynamics of miR-1260b in MFS cells, we investigated the intracellular expression level of Heatmap and the hierarchical clustering of miRNA microarray analysis using serum samples collected from MFS patients in preoperative state (MFS), age-matched benign tumor patients (NON), and healthy individuals (CONT). (B) Serum levels of four upregulated miRNAs in (A). *p < 0.05, **p < 0.01; Student's t test. (C) The expression levels of extracellular miR-642a, miR-1260b, and miR-4286 in MFS cell culture media (NMFH-1 and NMFH-2) compared to human mesenchymal stem cells (hMSCs). Data are presented as mean ± S.D. (n = 3 in each group) *p < 0.05, **p < 0.01; Student's t test. (D) The expression levels of extracellular miR-1260b in the culture media of NMFH-1 and NMFH-2 cells based on cell number and the culture duration. Data are presented as mean ± S.D. (n = 3 in each group) *p < 0.05; Student's t test.    www.nature.com/scientificreports www.nature.com/scientificreports/ miR-1260b in the tumor-derived EVs secreted by the MFS cells. The intracellular expression levels of miR-1260b were not upregulated in the MFS cells compared to those in the control hMSC cells ( Supplementary Fig. 2). Next, we characterized the purified EVs from the culture media of both NMFH-1 and NMFH-2 cells. Negative-stain transmission electron microscopy (TEM) revealed that the isolated particles were essentially homogeneous vesicles approximately 100 to 150 nm in diameter (Fig. 3A). Dynamic light-scattering analysis using Zetasizer Nano-S confirmed the presence of vesicles with a peak size of 109 nm and 87 nm in NMFH-1 and NMFH-2 cells, respectively (Fig. 3B). Western blotting of the EV fractions confirmed the expression of the tetraspanin protein, CD63, which is a known EV marker. However, only whole-cell lysates showed the expression of cytochrome-C (Cyt-C), a mitochondrial marker, and calnexin, an endoplasmic reticulum (ER) marker, supporting the quality of EVs preparations (Fig. 3C, Supplementary Fig. 3). In contrast, the whole-cell lysates and EVs showed the expression of vimentin, a mesenchymal marker. Additionally, the whole-cell lysates showed the expression of tubulin (Fig. 3C, Supplementary Fig. 3). We then investigated the expression level of miR-1260b in the EV fractions and observed significant elevated expression compared with the EVs purified from control hMSCs (Fig. 3D). We also checked the expression level of miR-1260b after depletion of EVs from MFS cells using cambinol, an inhibitor of neutral sphingomyelinase-2, which plays a regulatory function in EVs budding from the plasma membrane 38,39 . Cambinol treatment significantly decreased extracellular miR-1260 levels in both NMFH-1 and NMFH-2 cells (Fig. 3E). In addition, we treated secreted EVs with RNase before qRT-PCR to confirm that miR-1260b is embedded in EVs. No significant differences in the RNA amount and miR-1260b and cel-miR-39 expression levels were www.nature.com/scientificreports www.nature.com/scientificreports/ observed between the RNase A treated group and the control group (Fig. 3F). The results indicated that the extracellular miR-1260b was embedded in the MFS-derived EVs. Furthermore, we isolated EVs from serum samples obtained from MFS patients and healthy individuals in order to investigate miR-1260b expression levels from the EVs. miR-1260b expression in the patient-derived EVs was also amplified, and was found to be significantly upregulated in the EVs from MFS patients compared with healthy individuals (Supplementary Fig. 4).

Enhanced infiltration of MFS by EV-miR-1260b via stimulation of normal fibroblasts (NFs).
To further investigate the functional significance of extracellular miR-1260b, we performed an invasion assay using MFS cells and NFs, educated using tumor-derived EVs and synthetic miR-1260b. We established an in vitro model mimicking the tumor microenvironment using MFS cells, NFs, and matrigel chamber. NFs, which had been transfected with synthetic miR-1260b or educated using tumor-derived EVs, were plated onto growth factor reduced matrigel (Fig. 4A). We first labeled MFS-derived EVs with PKH67 and cultured them with NFs, which enabled us to detect tumor-derived EVs taken up by NFs (Fig. 4B). We then analyzed the expression levels of miR-1260b in NFs educated using NMFH1-and NMFH2-derived EVs. miR-1260b expression levels were significantly higher in the EV-educated control NFs (Fig. 4C). In addition, the education of the EVs on NFs significantly stimulated the invasion of MFS cells (Fig. 4D). We further assessed the cellular invasiveness of MFS using miR-1260b overexpressed NFs. First, we confirmed that miR-1260b were overexpressed in the NFs with synthetic miR-1260b transfection compared to miR-NC (Fig. 4E). We then observed that cellular invasiveness in NMFH-1 and NMFH-2 was enhanced in the co-culture environment with miR-1260b overexpressed NFs (Fig. 4F). The results suggested that EV-miR-1260b promoted cellular invasion of MFS via the stimulation of NFs. Finally, we assessed the cellular invasiveness of miR-1260b overexpressed MFS without NFs (Fig. 4G). The numbers of invaded NMFH-1 and NMFH-2 cells, overexpressing miR-1260b, were not enhanced without in the absence of NFs (Fig. 4H). We also evaluated the effect of EVs from NFs in MFS cells in terms of cellular invasion. The cellular invasion of MFS cells did not change with or without education of EVs from NFs, indicating the EVs from NFs have minimal effect on MFS invasion ( Supplementary Fig. 5). Collectively, the results above demonstrate that EV-miR-1260b induces MFS cell invasion not through "autocrine" but through "paracrine" mechanisms from MFS cells, communicating with the adjacent normal mesenchymal cells.
PCDH9 as a direct target of miR-1260b in fibroblasts regulating MFS infiltration. To identify target genes of miR-1260b, we carried out mRNA profiling using two different microarray analyses. Twenty-five genes were identified as decreased mRNAs with at least two-fold changes in the miR-1260b overexpressed NFs compared to the control NFs (miR-NC; Supplementary Table 7). The second microarray was performed using RNA collected using anti-Argonaute 2 antibody immunoprecipitation (Ago-2 IP) in NF cells transfected with miR-1260b or miR-NC. A total of 294 genes were identified as upregulated mRNAs with at least four-fold changes in the former samples (Supplementary Table 8). The combined data identified 15 candidate genes as miR-1260b targets (Fig. 5A, Supplementary Table 9). Among the 15 candidate genes, qRT-PCR revealed that PCDH9 was significantly downregulated in NFs with miR-1260b overexpression (Fig. 5B) or treatment with EVs derived from MFS cells (Fig. 5C). Finally, we investigated whether the targets were correlated with cellular invasiveness of MFS through the stimulation of NFs. Silencing of PCDH9 in the NFs promoted cellular invasion of MFS in the co-culture environment of MFS cells and NFs (Fig. 5D,E). The results indicated that tumor-derived EV-miR-1260b promote local aggressiveness of MFS via the downregulation of PCDH9 in the peri-tumoral normal mesenchymal cells.
In vivo dynamics of serum miR-1260b expression levels in MFS-bearing mice. Finally, we established an MFS xenograft mouse model using the NMFH-1 cell line and investigated the correlations between tumor growth and serum miR-1260b levels in order to evaluate the in vivo dynamics of serum miR-1260b expression levels (Fig. 6A). Along with increased tumor size (Fig. 6B), serum miR-1260b expression levels were elevated significantly in MFS-bearing mice (Fig. 6C). In addition, we observed a significant correlation between tumor volume and serum miR-1260b expression levels (R = 0.7253, p = 0.005; Fig. 6D). These results, together with the results using patients' serum (Fig. 1B, Fig. 2A-C), suggested that serum expression of miR-1260b could reflect tumor burden in MFS-bearing animals and humans.

Discussion
Emerging evidence has demonstrated the potential clinical utility of circulating miRNAs as diagnostic and prognostic biomarkers in malignant diseases. Several studies have demonstrated the potential in various types of cancers including lung [40][41][42] , breast [43][44][45] , prostate 46,47 , colorectal 48,49 , gastric [50][51][52] and liver cancers [53][54][55] . However, the evidence for STSs is limited to synovial sarcoma 30,56 , rhabdomyosarcoma 57 , and MPNST 58 . In addition, the functional relevance of circulating miRNAs with tumor progression is rarely discussed in sarcomas. To the best our knowledge, the present study is the first to describe the global profiling of circulating miRNAs in patients with MFS and to demonstrate their functional role in local MFS aggressiveness.
We observed dissimilar miRNA dysregulation patterns between intracellular and extracellular miRNA expression. For example, three miRNAs (miR-642a, miR-1260b, and miR-4286) were upregulated in MFS cell culture media compared to control hMSCs, while the cellular expression of the miRNAs was not upregulated in the MFS cells compared to the control hMSC cells. The dissimilarity between cellular miRNAs and secreted miRNAs are also observed in various cancers, such as breast cancer [59][60][61] , osteosarcoma 29 , and synovial sarcoma 30 . Although the reason for the dissimilarity remains unknown, the existence of molecular mechanisms regulating the secretion of miRNAs might exist. Therefore, our global miRNA profiling approach focusing on the extracellular/circulating miRNAs is potentially appropriate for the investigation of clinically useful miRNA candidates rather than focusing on intracellular miRNA dysregulation of malignant tumors. Although we did not identify any differences in www.nature.com/scientificreports www.nature.com/scientificreports/ miR-1260b expression between tumor cells/tissues and normal cells/tissues, differences were observed when we focused on circulating miRNAs, which supports their use as sources to look for new biomarkers for diagnosis or disease monitoring in a noninvasive way.
To date, the oncogenic role of miR-1260b has been reported in prostate cancer, renal cell carcinoma, non-small cell lung cancer, and colorectal cancer [62][63][64][65] . Silencing of miR-1260b in prostate cancer cells enhanced the expression of sFRP1 and Smad4 and decreased cellular proliferation, invasion, and migration 62 . In addition, miR-1260b, highly expressed in renal cell carcinoma, promoted cellular proliferation and invasion via the reduction of the expression of several tumor suppressor genes associated with Wnt signaling, such as sFRP1, Smad4, and Dkk2 63 . The overexpression of miR-1260b is reportedly correlated with the development of non-small cell lung cancer and prognosis of lymph node metastasis 64 . The studies above have demonstrated overexpression of intracellular miR-1260b, which indicate its oncogenic role in several cancer cells. However, the significance of extracellular/ circulating miR-1260b in malignant tumors remained unclear. Therefore, our investigation is the first to describe the significance of the extracellular/circulating miR-1260b, which could be a promising non-invasive biomarker for the monitoring of MFS.
Recent investigations have demonstrated that extracellular miRNAs stimulate tumor progression in the tumor microenvironment 37,66-68 . Tumor-derived EVs, containing nucleic acids or proteins, have been shown to play a crucial role in signal exchange with recipient cells [69][70][71][72] . For example, extracellular miR-150, secreted from monocytes, could enter human vascular endothelial cells and enhance cell migration of the recipient cells 67 . In another study, extracellular miR-9, derived from various tumor cells, were delivered to endothelial cells and mediated breast cancer cell invasion 73 . The breast cancer-derived miR-105 effectively reduced the expression of the junctional proteins, including ZO-1 in endothelial cells, and facilitated metastasis 37 . In the present study, extracellular miR-1260b, packaged in tumor-derived EVs facilitated cellular invasion of MFS cells, which was not through an "autocrine" but through "paracrine" mechanism by alternating miRNA and mRNA expression within the peri-tumoral fibroblast cells. The results indicate that tumor-derived EVs containing miRNAs of donor cells could The infiltrative growth of MFS, which is radiologically observed as a tail-like pattern under MRI, is associated with high local recurrence rates and subsequent decreased mortality 8 . In the present study, we revealed that miR-1260b-PCDH9 in the NFs are associated with cellular infiltration of MFS. Protocadherins (PCDHs) are a group of adhesion proteins that play an important role as regulators of cell-cell adhesion. PCDH9 is located at 13q21.32 chromosomal loci in humans and its dysregulated expression is associated with the malignant tumor phenotype 74 . PCDH9 is downregulated in glioblastoma and overexpression of PCDH9 inhibits cellular invasion of glioma 75,76 . Silencing of PCDH9 decreased the expression of epithelial markers, such as E-cadherin and Occludin 77 . The infiltrative nature of MFS could be explained by the inhibition of cell-cell adhesion between tumor and normal mesenchymal cells, which may be caused by PCDH9 suppression via tumor-derived EV-miR-1260b. In this in vitro model, the increased invasiveness of MFS cells could be related to not only pretreatment of EV-miR-1260 to NFs, which downregulated PCDH9 expression, but also excessive secretion of miR-1260b in the extracellular space from MFS cells affecting the NFs locally.
Another unique strength of our study is that we could establish an in vivo animal model for liquid biopsy and validate our results from human clinical samples with the use of our model. Although our methods of MFS xenograft model may not represent the complexity in patients, serum expression levels of miR-1260b, selected from the array analysis using serum samples from patients and culture media of MFS cells, was significantly correlated with tumor growth. A recent report from Hur et al. describes an in vivo metastasis animal model for colorectal cancer (CRC). In their model, serum miR-203 expression levels were significantly elevated in CRC-bearing mice with liver metastasis compared with nonmetastatic mice 78 . In addition, they reported a significantly positive correlation between serum miR-203 levels and tumor volume of metastatic nodules 78 . Although future studies are necessary to clarify whether serum miR-1260b levels could also be used as a tumor-monitoring marker for www.nature.com/scientificreports www.nature.com/scientificreports/ metastasis of MFS, we believe that our in vivo animal model contributes to the establishment of liquid biopsy in other sarcoma models.
In summary, our global miRNA profiling analysis revealed the upregulation of circulating/extracellular miR-1260b in patients with MFS. In addition, serum miR-1260b expression levels were correlated with tumor burden in both MFS-bearing mice and patients and radiological tail-like sign, which represent the infiltrative nature of the tumor. Furthermore, tumor infiltration was mediated by tumor-derived EV-miR-1260b, which decreased PCDH9 expression in peri-tumoral NFs cells. The above findings offer novel insights into the infiltrative nature of MFS based on cell-cell communication via tumor-derived EVs and demonstrate the potential of non-invasive blood-based monitoring in patients with the aggressive STS.

Materials and Methods
Patients and samples. The study protocol was approved by the Institutional Review Board. Written informed consent was obtained from all patients and healthy individuals involved in this study. Blood samples were collected from MFS patients (n = 15; Supplementary Table 1), age-matched nonsarcoma patients (n = 5; Supplementary Table 2), and healthy control individuals (n = 9; Supplementary Table 3) at our institute between 2013 and 2016. Among the samples collected from the 15 MFS patients, a total of five samples from five patients were used for microarray analyses and other samples (10 preoprerative and 10 postoperative) from 10 patients were used for tumor monitoring by the specific serum miRNA. Clinical data, including MRI patterns, age at surgery, gender, tumor site, tumor depth, tumor size, histological grade, presence or absence of recurrence, and presence or absence of metastasis were retrieved from the sarcoma database. Gadolinium-enhanced MRI (Gd-MRI) images were evaluated for signal characteristics and a tumor-infiltrative growth pattern. MFS exhibits an extensive multidirectional spread pattern, particularly along the fascial plane and this growth pattern is defined as a tail-like pattern 9,79 . In contrast, Gd-MRI images of some MFS cases do not exhibit extensive signal spread and good demarcation, and are described as a solid pattern 3 . Venous blood samples were preserved at 4 °C after collection from patients. Serum samples were separated from venous blood samples by centrifugation at 3,500 rpm for 15 min at 4 °C. The serum samples were then centrifuged at 20,000 × g for 15 min 4 °C to prevent contamination by cellular nucleic acids and then passed through a 0.22-µm pore filter (Merck Millipore, Billerica, MA., USA). The sera samples were stored at −80 °C until further processing.
Cell lines and cell culture. The human MFS cell lines, NMFH-1 and NMFH-2, were obtained from Niigata University and human mesenchymal stem cells (hMSCs) were obtained from Lonza (Walkersville, USA). Both the MFS cell lines were maintained in Roswell Park Memorial Institute media (RPMI, Gibco Laboratories, Grand Island, NY, USA.). Normal skeletal muscle tissues were obtained from 3 patients undergoing wide resection of MFS tumor. Normal skeletal muscle tissues placed and cultured on collagen coating dish in culture media which select for fibroblast growth Dulbecco's Modified Eagle's medium (DMEM) (NACALAI TESQUE, Inc., Kyoto, Japan) for 1 week, fibroblast cells were grown. Primary cultures of normal fibroblasts (NFs) were established.
All culture media were supplemented with 10% fetal bovine serum (FBS, Hyclone, Victoria, Australia), penicillin (100 U/mL), and streptomycin (100 mg/mL) (NACALAI TESQUE) in a humidified atmosphere containing 5% CO 2 at 37 °C. miRNA microarray analysis. The miRNA microarrays were manufactured by Agilent Technologies (Tokyo, Japan), and 2 ng of extracted RNA were used for each microarray experiment. RNA quantity and quality were determined using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Massachusetts, USA) and an Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), as previously described 80 . Total RNA was labeled with cyanine 3 (Cy3) using a miRNA Complete Labeling and Hyb Kit (Agilent Technologies) following the manufacturer's instructions. In brief, total RNA was dephosphorylated using Calf Intestinal Alkaline Phosphatase (CIP) Master Mix incubated at 37 °C for 30 min. Dephosphorylated RNA was denatured with DMSO incubated at 100 °C for 5 min and then immediately transferred to ice for 2 min. These products were mixed with a ligation master mix for T4 RNA ligase and Cy3-pCp (cyanine 3-cytidine biphosphate) and incubated at 16 °C for 2 hours. Labeled RNA was dried using a vacuum concentrator at 55 °C for 1.5 hours. Cy3-pCp-labeled RNA was hybridized on an Agilent SurePrint G3 Human miRNA 8 × 60 K Rel.19 (design ID: 046064) array at 55 °C for 20 hours. After washing, microarrays were scanned using an Agilent DNA microarray scanner. Intensity values of each scanned feature were quantified using Agilent Feature Extraction software ver. 10.7.3.1, which performs background subtractions. We only used features that were flagged as no errors (detected flags) and excluded features that were not positive, not significant, not uniform, not above background, saturated, and population outliers (not detected flags). The expression analysis was performed with Agilent GeneSpring GX software version 13.1. There are a total of 2,006 miRNA probes on SurePrint G3 Human miRNA 8 × 60 K Rel.19 (design ID: 046064) without control probes. Isolation of EVs from cell culture media. EVs were isolated from culture medium (CM) supernatant of MFS cells, hMSCs, and NFs. CM from cell lines that were grown to 50-70% confluence was exchanged with FBS free culture medium and was collected after a 24-h culture. The collected CM was first centrifuged at 3,500 rpm for 15 min and 9,000 × g for 30 min at 4 °C followed by filtration through a 0.22-μm filter (Merck Millipore) to remove cellular debris, unwanted proteins, and any live or dead cells. The supernatant was concentrated to about 1 ml using 100-kDa MWCO ultrafiltration membranes (Fisher Scientific, Loughborough, UK). Afterwards, the concentrated sample was ultracentrifuged at 100,000 g for 70  www.nature.com/scientificreports www.nature.com/scientificreports/ was disposed, and the EV pellet was rinsed with PBS at 100,000 × g for 70 min at 4 °C. The EV pellet was finally resuspended in PBS and its protein concentration was measured using Bradford Protein Assay Kit (Takara Bio, Shiga, Japan). The obtained EVs were confirmed under transmission electron microscopy (TEM) by the conventional negative staining procedure performed on two different samples. In brief, 10 μl aliquots of EV suspension were sedimented for 2 min onto a 300 mesh, copper/carbon-coated grid and then negatively stained with 2% uranyl acetate and observed with a TEM Hitachi H-7650 at 80 Kv, as previously described 81 . Light-scattering analyses were performed using a Zetasizer Nano-S (Malvern Instruments Ltd., Worcestershire, UK) and a 40-μL cuvette (Malvern Instruments). Samples were equilibrated for 2 min at 25 °C followed by three readings consisting of 10 measurements each. To deplete EVs from tumor cells, cambinol was purchased from Sigma-Aldrich (St. Louis, MO, USA). This was dissolved in dimethylsulfoxide (DMSO, Sigma-Aldrich) according to manufacturer's instructions. Cambinol dissolved in DMSO was used at 10 μM and the same concentration of DMSO was used for the control. MFS cells were treated with cambinol in serum-free media for 48 hours.
Isolation of EVs from human serum. EVs were purified from human serum samples by size exclusion chromatography on drip using EV-second (GL sciences, Tokyo, Japan) as previously described 30 . The column was initially equilibrated with 700 μl of PBS twice, followed by a blocking step using 700 μl of FBS. After repeating the wash steps six times with 700 μl of PBS, 200 μl of the collected human serum sample was loaded onto this column followed by collection of 12 consecutive fractions in 100 μl of PBS. CD9 expression in these fractions was measured by CD9 sandwich ELISA and CD9-positive fractions were recognized as the EV-rich portion.
RNase protection assay. To degrade unprotected RNA and the isolation of RNA only confined within EVs, purified EVs were treated with RNase A (0.5 mg/mL; Takara Bio) for 20 min at 37 °C, followed by enzyme inactivation. Untreated purified EVs were also tested as controls.

Reverse transcription (RT) and quantitative real-time PCR. Reverse transcription and quantitative
real-time PCR were performed, as previously described 30 . The isolated RNAs were reverse transcribed using the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). The cDNA products were mixed with 5.0 µL of TaqMan 2× Universal PCR Master mix and 0.50 µL of each primer for qPCR using Agilent Mx3000P (Agilent Technologies, Santa Clara, CA, USA) instrumentation. Data obtained from these procedures were analyzed using the 2 −∆∆Ct method 82 . The expression levels of miRNA were normalized using cel-miR-39 for serum and culture media, and RNU6B for tumor and other cells. Differences between the groups are presented as ΔCt, indicating differences between Ct values of miRNAs of interest and Ct values of normalizer miRNAs 30 .
Western blotting. Western blotting was performed as previously described 30 . Total protein from cells (10 μg) and exosomes (10 μg) was fractionated using an electrophoretic gradient across Mini-PROTEAN tris-glycine extended gels (BIO-RAD, Richmond, CA, USA). Measurement of protein concentration was performed using a Bradford Protein Assay Kit (Takara Bio), according to the manufacturer's instructions. The gels were then transferred onto Immun-Blot PVDF membranes (BIO-RAD) under wet electrophoretic conditions. The blotted protein was blocked for 1 hr at room temperature with Odyssey blocking buffer in PBS (LI-COR, Lincoln, NE, USA) and was followed by incubation overnight at 4 °C with the following primary antibodies: 1:1000 anti-CD63 mouse monoclonal antibody (BD Biosciences, San Jose, CA, USA); 1:1000 anti-cytochrome-c mouse monoclonal antibody (BD Biosciences); 1:1000 anti-calnexin rabbit monoclonal antibody (Abcam) and 1:1000 hFAB Rhodamine (#12004168 -Bio-Rad). Primary antibodies were detected by IRDye 800CW anti-rabbit IgG and IRDye 680RD anti-mouse IgG secondary antibodies (LI-COR) and were incubated with the protein-blotted membrane for 1 hr at room temperature. Fluorescence was then detected on the Odyssey imaging system (LI-COR).
Animal experiments. All the animal experiments were approved by the Institutional Animal Care and Use Committee of Okayama University and conducted in compliance with a protocol reviewed by the committee. BALB/c nu/nu female mice were purchased from CLEA Japan Inc. (Tokyo, Japan) at 4 weeks of age. Twelve mice, four mice per group, were inoculated subcutaneously with NMFH-1 cells at a concentration of 5 × 10 6 cells in 100 μl of RPMI on the right buttock under general anesthesia performed with 2% isoflurane. Four control mice received 100 μl RPMI by the same route. Tumor sizes were measured after one week of injection and then measured once a week. Blood samples were obtained by cardiac puncture at three, five, and seven weeks after injection under general anesthesia. The animals were sacrificed after blood sampling. The blood samples were stored at 4 °C in CAPIJECT micro collection tubes (TERUMO, Tokyo, Japan), and blood processing was done within 2 h of sampling as previously described for patient serum samples. miRNA and siRNA transfection. Synthetic miRNA mimics (Mimic hsa-miR-1260b and negative control (NC); Gene Design, Ibaraki, Japan) were transfected into Normal Fibroblast (NF) cells at 30 nM using DharmaFECT 1 (Dharmacon, Tokyo, Japan). Twenty-four hours after transfection, NF cells were reseeded. For RNAi experiments, NF cells were transfected with Silencer Select Pre-designed siRNA (Thermo Fisher Scientific) using DharmaFECT 1 (Dharmacon) according to the manufacturer's protocols.
Cell invasion assay. NFs (2 × 10 4 cells) were plated onto 25 μg Matrigel-coated 24-well FluoroBlock cell culture inserts with 8-μm pore sizes (Corning Life Sciences, Durham, NC). Twenty-four hours after NF cells were transfected with miR-1260b mimic or miRNA mimic control were cultured, PKH67 (Sigma-Aldrich, St. Louis, MO, USA) -labeled NMFH-1 and NMFH-2 cells in RPMI with 1% bovine serum albumin (Sigma) were incubated on the monolayers of NF cells and allowed to invade for 24 h. Culture medium of the lower chamber