Human dermis microvascular endothelial cells (HMVECs) and human umbilical vein endothelial cells (HUVECs) were obtained from Lonza (Basel, Switzerland). Immortalized HMVECs (iHMVECs) and immortalized HUVECs (iHUVECs) were established by transducting SV40 large T antigen and hTERT as previously described (17). Briefly, lentiviruses were produced using HEK293T cells co-transfected with packaging constructs pCAG-HIVgp and the VSV-G- and REV-expressing construct pCMV-VSV-G-RSV-REV, pCSII-CMV-hTERT-IRES2-Venus, or pRRLsin-SV40 T antigen-IRES-mCherry. Each EC was analyzed and sorted by flow cytometry (FACSAria II) with Venus and mCherry signals to purify transducted cells. iHMVECs were cultured in the EC growth medium for microvascular cells (EGM-2MV) (Lonza), whereas HUVECs and iHUVECs were cultured in the EC growth medium (EGM-2) (Lonza). A375 cells were obtained from American Type Culture Collection (Manassas, VA, USA). A375SM cells (super-metastatic human melanoma cells) were kindly supplied by Dr. Fidler (M.D. Anderson Cancer Center, Houston, TX, USA) (18). The miRZip-1246 Anti-miR-1246 microRNA construct and Scramble Hairpin Control Anti-MicroRNA Construct (System Biosciences, CA, USA) were used to establish the Anti-miR-1246 A375SM and Control-miR A375SM cell lines, respectively. For the establishment of tdTomato-luc2 A375 and tdTomato-luc2 A375SM cell, ptdTomato-C1 vector (Clontech, Clontech, Palo Alto, CA, USA) and pGL4.50 [luc2/CMV/Hygro] vector (Promega, Madison, WI, USA) were inserted into the EcoRI-Notl region of pCSII-CMV-MCS to create pCSII-CMV-tdTomato-luc2 vector. Each lentiviral vector was transfected into HEK293T by FuGENE HD Transfection Reagent, along with the packaging vector pCAG-HIVgp and the VSV-G- and REV-expressing construct pCMV-VSV-G-RSV-REV (from H. Miyoshi, Department of Physiology, Keio University School of Medicine, Tokyo, Japan) according to the manufacturer’s instructions. A375 and A375SM cells transduced by each construct described above were cultured in minimum essential medium (GIBCO, Grand Island, NY, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS). All cells were cultured at 37°C in a humidified atmosphere containing 5% CO₂ and 95% air.

Six-week-old female nude mice (BALB/c AJcl-nu/nu; Clea, Japan) were housed under specific pathogen-free conditions. All animal care and experimentation procedures adhered to institutional guidelines and were approved by the Ethical Committee for Experimental Animal Care of Hokkaido University.

Tumor cells (1 × 10⁶ cells) were cultured for 48 h in each medium with 10% EV-depleted FBS. EV-depleted FBS was prepared by ultracentrifugation at 100,000×g for 16 h at 4°C (Optima XPN-80 and SW32Ti; Beckman Coulter, Miami, FL, USA). The medium was collected and centrifuged at 2,000×g for 10 min at 4°C, and the supernatant was filtered with a 0.2-µm filter unit (Nalgene; Thermo Scientific, MA, USA) to remove the debris. The filtered medium was ultracentrifuged at 175,000×g for 84 min at 4°C (Optima XPN-80 and SW32Ti) to isolate the EVs. The pellet was washed with PBS and ultracentrifuged under the same condition, and EVs were eluted in PBS. For EV preparation from serum, the ultracentrifugation was conducted at 210,000×g for 40 min at 4°C (Optima XPN-80 and SW55Ti). The micro BCA protein assay kit (Thermo Scientific) was used to estimate the protein concentration. For EVs treatment, 3 µg of EVs was used per 1.0 × 10⁵ cells unless otherwise stated. The same amount of PBS was used as the control.

The observation of A375SM-EVs by transmission electron microscope (JEM-1400; JEOL, Tokyo, Japan) was performed as previously described (15) ( Figure S1A ). The size of EVs was measured using Dynamic Light Scattering with a Zetasizer Nano ZS (Malvern Instruments, Southborough, UK) ( Figure S1B ). Western blotting was used with anti-HSP70 (1:1,000, BD Pharmingen, USA, #556433) and anti-Cytochrome C (1:500, BD Bioscience, #610607) antibodies. A375SM-EVs expressed the EV marker HSP70, but not cytochrome C, a cytoplasm protein ( Figure S1C ). Our experimental data have been submitted to EV-TRACK knowledgebase (EV-TRACK ID: 220296).

Total RNA was isolated with ReliaPrep RNA Cell Miniprep System (Promega). As previously described, first-strand cDNA was synthesized using ReverTra-Plus (Toyobo, Osaka, Japan) (19). Real-time qRT-PCR was performed using the KAPA SYBR FAST qPCR Kit (KAPA Biosystems, Boston, MA, USA) according to the manufacturer’s instructions. Cycling conditions were set based on CFX Manager (Bio-Rad, Hercules, CA, USA). Relative mRNA expression levels were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and analyzed using the delta-delta-Ct method. The primer sequences were as follows:

human GAPDH: forward, 5′-ACAGTCAGCCGCATCTTCTT-3′; reverse, 5′-GCCCAATACGACCAAATCC-3′, human ICAM-1: forward, 5′-GGCAAGAACCTTACCCTACGCTGCC-3′; reverse, 5′-GTTCAGTGCGGCACGAGAAATTGGC-3′, human VE-Cadherin: forward, 5′-AAGTACAGCATCTTGCGGGGCGAC-3′; reverse, 5′-TTGATGATGCCCTCGTTGTGGGCG-3′, human Claudin-5: forward, 5′-AAGATTGAGAGCTGCCAGAGGC-3′; reverse, 5′-TACCCTCTTTGAAGGTTCGGGG-3′, human ZO-1: forward, 5′-GGGGAGGGTGAAGTGAAGA-3′; reverse, 5′-AGGCATTTCTGCTGGTTAGTATG-3.

Total RNA was isolated from the cells or EVs using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and RNeasy Mini Kit (Qiagen, Valencia, CA, USA). Cel-miR-39 (Hokkaido system science, Japan) was added (final concentration: 1 nM) to EVs solution. Reverse transcription was performed using the Taqman miRNA reverse transcription kit (Applied Biosystems, Carlsbad, CA, USA). miRNA expression levels were examined by qRT-PCR using Universal PCR Master Mix II (Applied Biosystems) and TaqMan MicroRNA Assays (Applied Biosystems) according to the manufacturer’s instructions. Cycling conditions were set based on CFX Manager (Bio-Rad). Relative miRNA expression levels were normalized to cel-miR-39 as external control. Primers were defined as follows:

miR-1246 (Assay ID: CSN1EFS), cel-miR-39 (Assay ID: 000200; UCACCGGGUGUAAAUCAGCUUG).

ECs were seeded at 1 × 10⁵ cells/well (6-well plate), followed by transfection with 50 nM miRIDIAN microRNA Human hsa-miR-1246-Mimic (CN-001040, Dharmacon; GE Dharmacon, Lafayette, CO, USA) using Lipofectamine RNAiMAX (Invitrogen) to a final concentration of 25 nM. miRIDIAN microRNA Mimic Negative Control (CN-001000-01-05; Dharmacon) was used as a negative control. After 6 h, the medium was changed and used for subsequent analysis.

miR-1246 mimic transfected iHMVECs were fixed with 4% paraformaldehyde (PFA) at room temperature for 10 min. After blocking with 5% goat serum in PBS for 1 h, the cells were incubated with mouse anti-human CD144 antibody (1:400, 555661; BD Pharmingen) at 4°C overnight. After washing with PBS, cells were incubated with Alexa Flour 594 goat anti-mouse antibody (1:400, 11032; Invitrogen) at room temperature for 2 h. Counterstaining was performed using DAPI (Dojin, Kumamoto, Japan) and mounted with Perma Fluor (Thermo Scientific). The stained sample was observed using a confocal microscope (FV10i; Olympus). Fluoview FV10-ASM Viewer Software (Olympus) was used for image processing. The stained area was calculated using ImageJ (National Institutes of Health, Bethesda, MD, USA) from 8 randomly selected fields of view.

iHMVECs with miR-1246 mimic transfection or EV treatment (10 µg) were incubated for 48 h on a 35-mm dish to form a monolayer. tdTomato-luc2 A375 or A375SM cells (2.0 × 10⁵ cells) were added to iHMVEC monolayers and incubated for 3 h. After removing non-adherent tumor cells by washing with PBS three times, cells were fixed with 4% PFA. Adherent tumor cells were observed with a confocal microscope (FV10i). Fluoview FV10-ASM Viewer Software was used for image processing. The number of adherent tumor cells was counted from 5 randomly selected fields using ImageJ.

iHUVECs were seeded at 5.0 × 10⁴ cells on 0.4 µm pore size Transwell filters (Corning, 3413, NY, USA) and incubated for 72 h to allow monolayer formation. After EVs treatment (3 µg), FITC-dextran (Sigma-Aldrich, FD70S) was added to the top well at 1 mg/ml and incubated for 30 min at room temperature. Then, 100 µl of the medium in the bottom well was transferred into a Nunc^® MicroWell 96-Well Optical-Bottom Plates with Polymer Base (Thermo Scientific, 165305). The fluorescence intensity was measured with a Varioskan Flash (Thermo Scientific).

The cells were lysed and collected in radioimmunoprecipitation assay buffer (Cell Signaling Technology, Beverly, MA, USA). The total protein concentration was determined using the BCA Protein Assay Kit (Pierce, Rockford, IL, USA). An equal amount of proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. The membrane was blocked with 5% skim milk/TBST for 60 min. The antibodies used are specific for STAT3 (1:1,000, Cell Signaling Technology, 12640S, Beverly, MA, USA), pSTAT3 (1:1,000, Cell Signaling Technology, 9145S), VE-Cadherin (1:1,000, Abcam, ab33168, Cambridge, UK), ICAM-1 (1:2,000, Abcam, ab53013), and β-actin (1:5,000, Cell Signaling Technology, 4970), with anti-rabbit HRP-labeled antibody as the secondary antibody (1:5,000, Cell Signaling Technology, 7074). The signals were developed using ECL Western Blotting Detection Reagent (GE Healthcare, Little Chalfont, UK) and detected using LAS-4000 mini image analyzer (FUJIFILM, Tokyo, Japan). The levels of pSTAT3, VE-Cadherin and ICAM-1 were normalized to that of β-actin by densitometry using Image J.

Each vector plasmid combined with miRNA mimic transfection was performed with DharmaFECT Duo reagent (Horizon Discovery), in accordance with the manufacturer’s protocol and then, HEK293 cells were collected 48 h after co-transfection with the 3’UTR assay vector (pmiR-CDH5) and miR-1246 mimic (Dharmacon), and luciferase activity was measured using the the Dual-Glo Luciferase Assay System (Promega) in accordance with the manufacturer’s protocol. Firefly luciferase activity was normalized to Renilla luciferase activity.

A 822 bp fragment from the 3’UTR of TET2 containing the predicted target sequence of miR-1246 (located at positions 924-930 of the TET2 3’UTR) were PCR-cloned from 3’ UTR clone of CDH5 for miRNA target validation (Origene, Rockville, MD, USA). Three prime A-overhang was added to the PCR products after 15 minutes of regular Taq polymerase treatment at 72˚C. The PCR products were cloned into a pGEM-T easy vector (Promega). The amplified products were ligated into the NheI and SalI sites of the 3’UTR of the firefly luciferase gene in the pmirGLO Dual-Luciferase miRNA Target Expression vector (Promega) to generate pmiR-CDH5. Primer sequences are as follows (shown 5’ to 3’): CDH5_F, CAGCTAGCTTCTCTGGAGAAGGCCTGGAAG and CDH5_R, CAGTCGACTATTGCCCAGGCTAAAGATTTT. Site-directed mutagenesis was performed in the seed sequences of CDH5. PrimeStar Max DNA Polymerase (Takara, Japan) was used for PCR amplification. Forward primer and reverse primer sequences are as follows (shown 5’ to 3’): CDH_Mut_F, CGCCTAACGAAGCTCTCTTTCTTTTCTCT and CDH5_Mut_R, GAGCTTCGTTAGGCGACCAGGTGAGGCAG.

Millicell hanging cell culture inserts with a pore size of 0.4 µm (MCHT24H48; Millipore, Bedford, MA, USA) were set on a 24-well plate, and HUVECs were seeded at 4 × 10⁴ cells on the insert. The medium-only wells were prepared for correction. The medium was changed after 48 h. A375SM-EVs treatment (3 µg) or miR-1246 transfection was performed 72 h after seeding, followed by TEER measurement. TEER was measured 24 h after its treatment. Millicell ERS-2 (Millipore) was used to measure TEER. The formula of TEER value was calculated as: (resistance of experimental wells − resistance of empty wells) Ω × 0.33 cm² (the membrane area of the cell culture insert).

2 × 10⁵ iHMVECs were seeded in 6-well plates, followed by 24 h starvation with 5%FBS-EBM2. ECs were pretreated with STAT3 phosphorylation inhibitor (S3I-201, Calbiochem) for 2 h. Then, ECs were treated with A375SM-EV (2μg) and S3I-201 (10 or 50 μM, vehicle: DMSO). These concentrations were based on previous reports (20, 21). Total RNA was extracted 12 h later for further mRNA expression analysis.

In the experimental group (EV injection), A375SM-EV (3 µg) was injected intravenously to mice twice a week (five times in total), and then tdTomato-luc2 A375 (2.0 × 10⁵ cells) was injected intravenously. In the control group, PBS was used instead of EVs. After 24 h, VivoGlo Luciferin, In Vivo Grade (Promega), was intraperitoneally administered, and the tumor cell signals were detected using IVIS Spectrum (Caliper Life Sciences, Hopkinton, MA, USA) to analyze tumor cell adhesion to the lung. Forty days after injection, tumor cell signals were measured to analyze lung metastasis. The excised lungs were stained with H-E staining, and the number of metastatic lesions was counted.

Anti-miR-1246 A375SM cells (1.0 × 10⁶) or Control-miR A375SM cells (1.0 × 10⁶) were suspended in HBSS and subcutaneously injected into the right flanks of mice to analyze the effect of miR-1246 in tumor metastasis. The serum was used to analyze the EV-miR-1246 level after 28 days, the lungs were removed, and the GFP signals expressed in tumor cells were evaluated using IVIS Spectrum.

Anti-miR-1246 A375SM cells (1.0 × 10⁶) or Control-miR A375SM cells (1.0 × 10⁶) were suspended in HBSS and subcutaneously injected into the right flank of mice. After 4 weeks, Rhodamine-dextran (MW 70,000, Invitrogen, D1841) was administered intravenously 3 h before transcardiac perfusion. The tumors removed were sectioned after making a frozen block using the Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA, USA). The frozen sections were stained with DAPI for 10 min and then observed using a BZ-X810 microscope (Keyence Corporation, Itasca, IL, USA). Rhodamine signal was quantified using BZ-X800 Analyzer software (Keyence).

Statistical analysis was performed using JMP version 13 (SAS Institute). All the data are presented as mean ± standard deviation unless otherwise stated. The Student’s t-test or Wilcoxon test was used for comparing the two groups. Differences among groups were determined using one-way ANOVA, followed by a Tukey–Kramer multiple comparison test. A P-value <0.05 was considered to be significant.

We analyzed the effects of tumor EVs on ECs in lung metastasis using a high metastatic tumor cell line, A375SM. We have previously reported the differential effects of tumor EVs on ECs, and the ECs’ phenotype is altered to a drug-resistant or proangiogenic phenotype (8, 15). It has been reported that tumor EVs induced metastasis (14, 22).

The tumor cell adhesion to the lung and lung metastasis was analyzed to address the alteration of the metastatic potential of a low metastatic tumor cell line, A375, after pretreating EVs isolated from A375SM ( Figure 1A ). The A375SM-EV treated group had more signals of tumor cells in the lung than the control group 24 h after tumor cell injection ( Figures 1B, C ), suggesting that tumor adhesion to the lung was stimulated by pretreatment of A375SM-EV. Similarly, A375SM-EV treated mice showed increased lung metastasis 40 days after tumor cell injection ( Figures 1D, E ). The A375SM-EV group showed more metastasis nodules in the lung than the control group in the histological study ( Figures 1F, G ). The above results suggest that EVs from high metastatic tumors induce tumor cell adhesion to the lung and promote lung metastasis.

We have previously reported that miR-1246 levels are higher in A375SM-derived EVs than in A375 (15). To evaluate the role of miR-1246, we established a stable miR-1246-knockdown A375SM (Anti-miR-1246 A375SM) using a lentivirus vector. miR-1246 levels in anti-miR-1246 A375SM were downregulated in cells and EVs compared to control-miR A375SM ( Figure S2 ). Anti-miR-1246 A375SM and Control-miR A375SM were injected into mice subcutaneously, and their tumors and lungs were removed after 28 days ( Figure 2A ). EVs were isolated from the serum of each tumor-bearing mice, and the miR-1246 levels were significantly lower in anti-miR-1246 A375SM tumor-bearing mice than in control-miR A375SM bearing mice ( Figure 2B ). There was no significant difference in tumor growth between the two groups ( Figure 2C ). However, the GFP signal showed that the lung metastasis was significantly suppressed in Anti-miR-1246 A375SM tumor-bearing mice compared to the control-miR A375SM ( Figures 2D, E ). These findings suggest that miR-1246, which is found in EVs of the high metastatic tumor, A375SM, promotes metastasis but not tumor growth.

We studied the adhesion of tumor cells to ECs as one of the mechanisms for metastasis by miR-1246. ICAM-1 expressed on ECs is the key molecule for tumor cells to adhere to ECs (23) and is induced via STAT3 activation (24).

A375SM-EV treatment upregulated mRNA and protein expressions of ICAM-1 in ECs ( Figures 3A, B ); similarly, miR-1246 increased ICAM-1 expression level ( Figures 3C, D ). On the contrary, anti-miR-1246 A375SM-EVs decreased the expression level of ICAM-1 in ECs ( Figure 3E ).

Furthermore, A375SM-EV treatment and miR-1246 transfection activated STAT3 in ECs ( Figures 3F, G ). STAT3 inhibitor cancelled the induction of ICAM-1 by A375SM-EV treatment ( Figure S3 ). These results demonstrated that miR-1246 in high metastatic tumor EVs increased ICAM-1 expression via the activation of STAT3. An adhesion assay was performed to investigate the effect of miR-1246 in EVs on tumor cells ( Figure 3H ). Both A375SM-EV treatment and miR-1246 transfection promoted A375 adhesion to ECs ( Figures 3I–L ). On the other hand, the number of A375 cells that adhered to ECs decreased upon anti-miR-1246 A375SM-EVs treatment ( Figures 3M, N ). In addition, even when high metastatic A375SM cells were seeded on the EC monolayer, both A375SM-EV treatment and miR-1246 transfection promoted their adhesion to ECs ( Figures S4A–D ). On the contrary, anti-miR-1246 A375SM-EVs treatment reduced A375SM adhesion to ECs ( Figures S4E, F ). These findings suggest that miR-1246 in high metastatic tumor EVs increases adhesion between tumor cells and ECs via ICAM-1 induction.

We next focused on vascular barrier dysfunction as a mechanism for promoting metastasis by miR-1246, and the cell–cell adhesion of ECs was analyzed. EC adhesion is composed of adherence and tight junctions (25). VE-Cadherin constitutes adherens junction, whereas tight junction consists mainly of claudin-5 and zonula occludens-1 (ZO-1) (26). Disrupting the EC barrier enhances vascular permeability and promotes metastasis.

miR-1246 transfection into ECs suppressed VE-Cadherin mRNA expression but not claudin-5 or ZO-1 ( Figure 4A ). We predicted the miR-1246 target using the publicly available database for miRNA target prediction (TargetScanHuman) and found that miR-1246 may bind to the 3′ UTR of VE-Cadherin ( Figure 4B ). The 3′ UTR reporter assay showed that luciferase activity tended to be decreased in VE-cadherin 3’ UTR construct, but not in the mutant. It suggested that VE-Cadherin could be either a direct target of miR-1246 or indirectly negatively regulated ( Figure 4C ). Moreover, western blotting revealed that miR-1246 transfection reduced VE-Cadherin protein expression in ECs ( Figure 4D ). Immunocytochemistry analysis of VE-Cadherin showed that miR-1246 attenuated the VE-Cadherin signals in ECs ( Figures 4E, F ). Next, we analyzed the change in endothelial permeability by adding A375SM-EVs to the ECs monolayer ( Figure 4G ). A375SM-EV treatment increased the dextran leakage to the lower chamber, indicating enhanced endothelial permeability ( Figure 4H ), and anti-miR-1246 A375SM-EVs attenuated the leakage induced by Control-miR A375SM-EVs ( Figure 4I ). Consistent with the above results, both A375SM-EV treatment and miR-1246 transfection reduced TEER, suggesting that miR-1246 containing tumor EVs dysregulated the EC barrier function ( Figures 4J, K ). Furthermore, we examined the vascular permeability by miR-1246 in vivo by dextran iv injection. Although there was no significant difference, the vascular permeability was found to be suppressed in tumors in anti-miR-1246 A375SM tumor-bearing mice ( Figures 4L, M ). Collectively, these results suggest that miR-1246 in high metastatic tumor EVs promotes tumor metastasis by disrupting EC barrier function via VE-Cadherin downregulation