Tang et al. BMC Genomics(2018) EARCH ARTICLEOpen AccessAlterations in exosomal miRNA profile uponepithelial-mesenchymal transition inhuman lung cancer cell linesYue-Ting Tang1,3†, Yi-Yao Huang1†, Jing-Huan Li4, Si-Hua Qin2, Yong Xu1, Tai-Xue An1, Chun-Chen Liu1,Qian Wang2*† and Lei Zheng1*†AbstractBackground: Epithelial–mesenchymal transition (EMT) is regarded as a critical event during tumor metastasis.Recent studies have revealed changes and the contributions of proteins in/on exosomes during EMT. Besidesproteins, microRNA (miRNA) is another important functional component of exosomes. We hypothesized that themiRNA profile of exosomes may change following EMT and these exosomal miRNAs may in return promote EMT,migration and invasion of cancer cells.Results: The small RNA profile of exosomes was altered following EMT. Kyoto Encyclopedia of Genes and Genomes(KEGG) pathway analysis revealed that the specific miRNAs of M-exosomes have the potential to drive signaltransduction networks in EMT and cancer progression. Co-culture experiments confirmed that M-exosomes canenter epithelial cells and promote migration, invasion and expression of mesenchymal markers in the recipient cells.Conclusion: Our results reveal changes in the function and miRNA profile of exosomes upon EMT. M-exosomescan promote transfer of the malignant (mesenchymal) phenotype to epithelial recipient cells. Further, the miRNAsspecifically expressed in M-exosomes are associated with EMT and metastasis, and may serve as new biomarkers forEMT-like processes in lung cancer.Keywords: Exosomes, Epithelial mesenchymal transition, miRNA, Lung cancer, Metastasis, High-throughput sequencingBackgroundLung cancer is one of the most common and lethal cancers worldwide [1], and metastasis is the leading causeof lung cancer-related deaths [2], which highlights theurgent need to better understand the critical steps andmechanisms of metastasis. Exosomes are small membrane vesicles (30–150 nm) secreted by a variety of cells.They are spherical particles enclosed by a phospholipidbilayer, containing DNA, RNA, and protein, and are released into the extracellular environment [3]. It has beenshown that exosomes are local and systemic cell-to-cell* Correspondence: [email protected]; [email protected]†Yue-Ting Tang, Yi-Yao Huang, Qian Wang and Lei Zheng contributedequally to this work.2Department of Clinical Laboratory, Zhujiang Hospital, Southern MedicalUniversity, Guangzhou, Guangdong, China1Department of Laboratory Medicine, Nanfang Hospital, Southern MedicalUniversity, No.1838 North Guangzhou Avenue, Guangzhou 510515,Guangdong, ChinaFull list of author information is available at the end of the articlemediators of information, through the horizontal transfer of signaling macromolecules, in various pathophysiological processes [4, 5]. Tumor cells may secrete largeamounts of exosomes, which are currently consideredone of the main contributors to tumor progression andmetastasis by promoting the proliferation and inhibitingapoptosis of tumor cells [6], activating angiogenic pathways [7], and enhancing invasiveness and migration oftumor cells [8, 9].Epithelial–mesenchymal transition (EMT), a process inwhich epithelial cells (E-cells) lose their polarity and areconverted into mesenchymal cells (M-cells), is regarded asa critical event during tumor metastasis [10, 11]. Severalstudies have revealed the contribution of exosomes duringthe EMT process in various cancers [12, 13]. On onehand, tumor-derived exosomes transfer specific cargoes torecipient cells, leading to the process of EMT and tumordevelopment [14–17]. On the other hand, the profiles of The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (, which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication o/1.0/) applies to the data made available in this article, unless otherwise stated.

Tang et al. BMC Genomics(2018) 19:802exosomal proteins and RNAs may also be altered following EMT [18–20]. Thus, the contents of exosomes may beused as new biomarkers to monitor the process of EMTand tumor development.Previous studies on exosomes have focused on the changesin the exosomal proteome during EMT [18, 20–22]. For example, the mRNA and protein levels of vimentin in exosomes were found to be increased after EMT [17]. However,microRNAs (miRNAs), another important functional component of exosomes, have rarely been reported in studies ofEMT in lung cancer. Only one study has reported that thelevel of miR-23a was increased in exosomes secreted by mesenchymal lung cancer cells [23]. Given that multiple specificmiRNAs may be secreted in tumor-derived exosomes andthat they may also be altered during tumor progression[24, 25], it is worth investigating the changes in theoverall exosomal miRNA profile during the EMTprocess. Hence, for the first time, we comprehensivelyanalyzed changes in the small RNA (sRNA) profiles ofexosomes following EMT of lung cancer cells usinghigh-throughput sequencing. Considering that miRNAstransferred by exosomes can alter the behavior of recipient tumor cells to promote oncogenesis [25, 26], wealso compared the function of exosomes derived fromE-cells with that of exosomes from M-cells carrying different sRNA cargoes, to gather evidence that reprogramming of the miRNA profile of exosomes mayaccelerate cancer malignancy.MethodsCell culture and establishment of EMT cell modelThe human lung cell line A549 and H1299 provided byCell bank, Type culture collection, Chinese Academy ofScience (CBTCCCAS) and human airway epithelial cellline 16HBE (kindly provided by the Chronic AirwaysDiseases Laboratory, Nanfang Hospital, Southern Medical University, Guangzhou) were cultured in RoswellPark Memorial Institute (RPMI-1640) medium, 10% fetalbovine serum (FBS) (System Biosciences, MountainView, CA, USA). Transforming growth factor-β1(TGF-β1) (R&D Systems, Minneapolis, Minnesota, USA)was dissolved with 4 mM HCL and diluted in phosphatebuffer solution (PBS) before use. A549 and H1299 cellswere incubated with fresh medium in the presence orabsence of TGF-β1 after 12 h starvation. The mostappropriate stimulating condition (concentration ofTGF-β1 (0, 2, 5 ng/ml) and duration of treatment (0, 24,48 h)) was confirmed by the result of functional studiesand evaluation of EMT-related markers. The A549 andH1299 cells untreated with TGF-β (PBS, 48 h) weredefined as E-phenotype cancer cells, and those treatedwith TGF-β (5 ng/ml TGF-β, 48 h) were defined asM-phenotype cancer cells.Page 2 of 14Cell culture medium (CCM) preparation and exosomesisolationWhen the same initial number (2 105/ml CCM) of E/Mphenotype A549, H1299 and 16HBE cells were grown to70–80% confluence, the FBS-containing media was removed and cells were cultured in serum-free 1640medium with 2% Exo-FBS exosome-depleted FBS (System Biosciences) for 48 h. Then the CCM samples(100 ml per sample), were collected and centrifuged at2000 g for 10 min and then filtered through 0.22-μmmembranes to remove dead cells, cell debris and largeparticles (shedding vesicles and apoptotic bodies).ExoQuick-TC (System Biosciences) was used for exosomes isolation, according to the manufacturer’s instructions. All centrifugations were performed at 4 C. Theexperiment was repeated three times using three completely independent sets of samples (three independentCCM samples prepared at different times). CON-exo,E1-exo, M1-exo, E2-exo, M2-exo represent exosomes derived from 16HBE, E-phenotype A549 cells, M-phenotypeA549 cells, E-phenotype H1299 cells, M-phenotypeH1299 cells, respectively.Nanoparticle tracking analysis (NTA)Exosome suspensions with concentrations between 1 107/ml and 1 109/ml were verified using a NanosightNS300 (Malvern, Great Malvern, UK) equipped with a405 nm laser to determine the size and quantity of particles isolated. A video of 60 s duration was taken with aframe rate of 30 frames/s, and particle movement wasanalyzed by NTA software (version 2.3, NanoSight).Transmission electron microscopy (TEM)Aliquots of 20–40 μl of a solution of exosomes were placedon a copper mesh and post-negatively stained with 2%phosphotungstic acid solution for 10 min. Subsequently,the samples were dried for 2 min under incandescent light.The copper mesh was observed and photographed under aHITACHI H-7650 transmission electron microscope(Hitachi High-Technologies, Tokyo, Japan).Western blot analysisExosomes or cell protein supernatants were denaturedin 5 SDS buffer and subjected to western blot analysis(10% SDS–polyacrylamide gel electrophoresis; 50 μgprotein per lane) using rabbit polyclonal antibodiesagainst E-cadherin, N-cadherin, vimentin (Cell Signaling,Danvers, MA, USA), CD9 and CD63 (Santa Cruz, CA,USA), TSG101 (Sigma, Dorset, UK) and calnexin (Bioworld Technology, MN, USA). The proteins were visualized on the Bio-Rad ChemiDoc XRS Imager system(Bio-Rad Laboratories, California, USA).

Tang et al. BMC Genomics(2018) 19:802Wound healing assaysCells were wounded using a 200-μl sterile pipette tip.Subsequently, the cells were washed twice with PBS andtreated with TGF-β1. The width of each wound wasmeasured and recorded 0, 24 and 48 h after thescratches were made.Page 3 of 14Exosomes labeling and tracking in A549, H1299 and16HBE cellsPurified exosomes were labeled with the PKH67 GreenFluorescent Labeling Kit (Sigma) according to the manufacturer’s recommendations. PKH67-stained exosomes wereincubated with recipient cells at 37 C for 24 h. The cellswere stained with DAPI (GeneCopoeia, Rockville, MD,USA) and visualized using an Olympus IX71 microscope.Migration and Matrigel invasion assaysThe Matrigel was uncoated (migration assay) or coated(invasion assay) on the upper surface of a transwellchamber (BD Biosciences, Franklin Lakes, New Jersey,USA), and 6 105 cells in serum-free medium containing TGF-β1 or exosomes were placed into the upperchamber. The chambers were then incubated in thelower chamber containing culture medium with 10%FBS for 24 h. The number of cells adhering to the lowermembrane was observed using an Olympus BX50 microscope (Tokyo, Japan) and digitized using ImageJ software(NIH Image).Isolation of exoRNA and cell RNA, and RNA analysisTRIzol-LS Reagent (Ambion, Life Technology, Carlsbad,CA, USA) was used to isolate high-quality total RNAfrom exosomes solution. The RNA concentration wasassessed using a Quibit 2.0 Fluorometer (Invitrogen, LifeTechnology, Carlsbad, CA, USA). The RNA yield andsize distribution were analyzed using an Agilent 2100Bioanalyzer with RNA 6000 Pico Kit (Agilent Technologies, Foster City, CA, USA). Cell RNA was extractedusing the TRIzol Reagent (TaKaRa, Dalian, China). ThemRNA level of EMT markers in the cells was analyzedusing a PrimeScriptTM quantitative real-time reagentKit (TaKaRa).Small RNA sequencingTo investigate differences in the miRNA profile amongexosomes from E-phenotype and M-phenotype A549cells, and 16HBE cells, sRNA high-throughput sequencingtechnology was used. The flowchart of sequencing grouppreparation is shown in Additional file 2: Figure S2A.Samples of 100 ng of 18 exoRNA or cell RNA were usedfor RNA library preparation, following the instructions ofthe TruSeq sRNA Sample Prep Kit (Illumina, San Diego,CA, USA). Subsequently, the PCR-amplified cDNA construct from 140 to 160 bp was purified. The purifiedcDNA was directly sequenced using an Illumina HiSeq2500 platform, and the 3′ adaptor sequences within theread sequences were cleaned up. Clean data were obtained by filtering out low-quality reads. All clean tagswere filtered and aligned with the NCBI GeneBank andmiRBase databases.Exosomes co-culture assayDifferent concentrations of exosomes (0, 50, 100 μg/ml)from E-phenotype or M-phenotype cells were co-culturedwith 16HBE, A549 and H1299 cells at 37 C for 48 h.After treatment, functional assays were conducted on recipient cells and the mRNA and protein from recipientcells were extracted and further investigated.Statistical analysisValues are expressed as mean standard deviation. All experiments were carried out in triplicate and repeated atleast twice. Student’s t-test and an Analysis of Variance(ANOVA) were used to determine the differences betweengroups. SPSS 15.0 was used for statistical analyses (SPSSIncorporated, Chicago, IL, USA), and P 0.05 was considered to be statistically significant.ResultsA549 cells undergo EMT after exposure to TGF-β1When the A549 cells were treated with TGF-β1 at differentconcentrations (0, 2, 5 ng/ml) for different durations (0, 24,48 h), the morphology, function, and expression of EMTmarkers of cells was changed. Morphologically, A549 cellsturned from being round into a spindle-like mesenchymalphenotype and lost intercellular junctions after TGF-β1treatment in a time- and concentration-dependent manner,which was most prominent after stimulation by 5 ng/mlTGF-β1 for 48 h (Fig. 1a). The process of EMT afterTGF-β1 stimulation was also accompanied by a decrease inknown epithelial marker (E-cadherin) and increasedmesenchymal marker (N-cadherin, vimentin, fibronectinand snail) levels (Fig. 1b, c). The mRNA level (Fig. 1b) andprotein level (Fig. 1c) of EMT-related markers changed in aconcentration-dependent manner but not a strict timedependent manner and the most significant change wasfound when stimulated with 5 ng/ml TGF-β1 for 48 h. Thewound healing and invasion assays showed that 5 ng/mlTGF-β1 induced EMT was associated with increased cellmigration (Fig. 1d) and invasion (Fig. 1e). In summary,A549 cells stimulated by 5 ng/ml TGF-β1 for 48 h showedthe most significant EMT characteristics, so this stimulationcondition was chosen to establish the EMT cell model. Inthe next study, A549 cells untreated with TGF-β1 were defined as E-phenotype cells (PBS, 48 h), and those treated

Tang et al. BMC Genomics(2018) 19:802Page 4 of 14Fig. 1 TGF-β1 was used to establish EMT cell models. a Morphology of A549 cells changed from E- to M- phenotype after TGF-β1 treatment. The mRNA(b) and protein (c) levels of EMT-related markers of A549 cells changed after being induced by TGF-β1. TGF-β1 significantly reduced E-phenotype marker(E-cadherin (E-cad)) levels, but increased the M-phenotype marker (N-cadherin (N-cad), vimentin (Vim), fibronectin (Fib) and snail) levels in a TGF-βconcentration-dependent manner but not a strict time-dependent manner. d The wound healing assays proved that TGF-β1 treatment can significantlyincrease cell migration abilities. The wound widths were significantly shorter at 24 h after TGF-β1 treatment than in the no-treatment group (P 0.05), andthis difference was more significant after 48 h treatment (P 0.01). e Invasion assays were used to determine cell invasion. Invaded cell numbers weresignificantly higher in TGF-β1 treatment than no-treatment groups, indicating TGF-β1 can improve cell invasion abilitywith TGF-β (5 ng/ml, 48 h) were defined as M-phenotypecells.significant differences in the number and average size ofexosomes derived from E-phenotype and M-phenotypeA549 cells (Additional file 1: Figure S1D).Biochemical characterization of exosomes from CCMTo demonstrate the presence of exosomes, the vesiclesisolated from A549 CCM were determined by TEM(Additional file 1: Figure S1A), western blotting(Additional file 1: Figure S1B) and NTA (Additional file 1:Figure S1C). A lipid bilayer structure around 100 nmobserved by TEM was consistent with descriptions of exosomes (Additional file 1: Figure S1A). The three exosomalmarker proteins (CD9, CD63 and TSG101) were presentin all vesicle samples (Additional file 1: Figure S1B). NTAdemonstrated that all particles were smaller than 300 nmand that most of them were about 50–200 nm in size(Additional file 1: Figure S1C). Additionally, there were nosRNA composition in cells differs from that in exosomesTo examine dynamic changes in the exosomal RNA(exoRNA) profile following EMT, high-throughput sequencing analysis was performed using RNA samples extractedfrom six groups: E-phenotype A549 cells (E-cells), exosomes derived from E-cells (E-exosomes), M-phenotypeA549 cells (M-cells), exosomes derived from M-cells(M-exosomes), 16HBE cells (Con-cells) and exosomes derived from 16HBE cells (Con-exosomes) (Additional file 2:Figure S2A). A progressive decrease of E-marker and increase of M-marker levels from Con-cells, E-cells toM-cells indicated an appropriate and dynamic EMT model

Tang et al. BMC Genomics(2018) 19:802(Additional file 2: Figure S2B). The majority of exoRNAwere small ( 200 nt), which differed from those of cells presenting specific ribosomal RNA (5S, 18S and 28S rRNA). Inaddition, there was variation in the RNA size-distributionbetween exosomes from A549 (E, M-exoRNA) and thosefrom 16HBE (Con-exoRNA) (Fig. 2a).Among the 18 samples, we obtained 228.28 millionclean reads. An average of 78.38% of exosomal cleanreads and 98.92% of cellular clean reads could bemapped to known RNAs of the human genome. Investigation of the chromosomal location of all the mappedsRNA revealed that the sRNA of exosomes mainlyoriginated from chromosome 1, which was significantlydifferent from the cell RNA location (mainly onchromosome 17). In addition to chromosome 1, manyPage 5 of 14sRNA of Con-exoRNA were also found on chromosomes 7 and 17 (Fig. 2b).The mappable sequences were annotated to miRNAand other small non-coding RNAs. The percentages ofmiRNA, rRNA, tRNA and other RNA were significantlydifferent between exoRNA and cell RNA: miRNA andtRNA were the most abundant known sRNAs in exosomes, while miRNA and rRNA were the most abundantsRNA in cells. In addition, cells contained a higher percentage (49.7–59.09%) of miRNA than exosomes(10.40–13.47%) (Fig. 2c). To better demonstrate the variation among the groups, an unsupervised hierarchicalclustering analysis was performed. As expected, the heatmap showed a clear separation between exoRNA andcell RNA. Either in exoRNA or in cell RNA, moreFig. 2 Comparison of small RNA profiles from cells and exosome extractions. a The length distribution of total RNA in cells and exosomes, asdetermined by Agilent RNA pico chip. (exo: exosomes for short) b All clean data was mapped by the National Center for Biotechnology Information(NCBI) to identify the chromosomal origin of small RNAs of interest. Outer rings display the ID of each chromosome, and the inner ring shows RNAabundance in the chromosomal region. c Pie chart summarizing the annotation of small RNA species. miRNA, rRNA and tRNA are the main identifiedRNAs in total distribution of small RNA. d Heatmap and cluster patterns of differentially expressed miRNAs among E/M/Con-exosomes and cells

Tang et al. BMC Genomics(2018) 19:802similar expression patterns were found between the Eand M groups than between the E and Con groups orthe M and Con groups (Fig. 2d).M-exosomes had distinct miRNA profiles which may beinvolved in EMT and cancer progression when comparedwith E-exosomes, con-exosomes or M-cellsThe contents of exosomes may change following EMT. First,we analyzed differences in the exosomal miRNA profilesamong three archetypical phases of the EMT process(Con-exosomes, E-exosomes, M-exosomes). Of all detectablemiRNAs, 264 miRNAs were common to E, M andCon-exosomes. A small amount of miRNAs were identifiedas unique for each group, such as hsa-miR-487b-3p, whichwas only detected in M-exosomes. (Fig. 3a). Regarding thedifferential level of miRNA expression among each group,50, 137 and 199 miRNAs were screened out betweenM-exosomes and E-exosomes, M-exosomes and Con-exosomes, E-exosomes and Con-exosomes respectively. 10 ofPage 6 of 14the most differentially expressed miRNAs were shown inTable 1. Additionally, the unique expressed miRNAs inM-exosomes compared with E-exosome, M-cell andCon-exosomes, were shown in Table 2.Exosomes are capable of altering the recipient cellphenotype by carrying RNA cargoes, such as miRNAs.Whether miRNA contained in M-exosomes can influence tumor development, including the EMT process,needs to be studied. To investigate the function of thesedifferentially expressed miRNAs, KEGG enrichmentpathway analysis was performed, including target genesof all the miRNAs that were differentially expressed between M-exosomes and E-exosomes, and the top 20 differentially expressed miRNAs between M-exosomes andCon-exosomes. The significant pathways are shown inFig. 3b and c. To our surprise, besides pathways involvedin exosome formation, transport, and other physiologicfunctions, the most enriched pathways were related totumor progression, such as classical signaling pathwaysFig. 3 miRNA profiles of M-exosomes showed functional enrichment of tumor EMT and progression. a Venn diagram comparing the number ofoverlapping miRNAs among E/M/Con-exosomes, or between M-exosomes and M-cells. b, c, d The KEGG analysis of the predicted targets ofdifferent miRNAs between groups (M- vs E-exosomes, M- vs Con-exosomes, M-exosomes vs M-cells). Red labels highlight the targets which wereenriched in pathway related to tumor progress

Tang et al. BMC Genomics(2018) 19:802Page 7 of 14Table 1 Top 10 significantly differentially expressed miRNAsbetween M-exo vs E-exo/Con-exo/M-cellTable 1 Top 10 significantly differentially expressed miRNAsbetween M-exo vs E-exo/Con-exo/M-cell (Continued)M-exo vs 69.671.66up5.397.01E-34miRNA IDM-exoM-cellup/down log2P-value(fold change) miRNA IDM-exoE-exoup/down log2P-value(fold change) o vs 43miRNA IDM-exoE-exoup/down log2P-value(fold change) 18.704.57E-21miR-143-3p476.7317 15.61803 M-cell vs E-cell4.9318931.04E-244miR-181a-2-3p 744down2.4793131.16E-21miR-145-5p15.63757 0up17.254661.05E-14miR-452138.8255123.3244 down1.6673826.41E-13miR-1180-3p119.0278 254.7887 down1.0980028.95E-11miR-494-3p47.59617 E-091.010528.64E-091.4649371.48E-08up20.77677 upmiR-365a-5p48.06983 23.8603miR-370-3p27.4212up9.933367 upM-exo vs Con-exomiRNA IDM-exoCon-exo up/down log2P-value(fold change) 3.671.45E-18The expression levels were showed by the number of reads per million cleantags (RPM). The mean values of three triplicate experiments showed in thetable. The edgeR bioconductor package was used to analyze the differencebetween groups and calculate the P values(ErbB, mTOR, FoXO, MAPK, PI3K-Akt, etc.), metabolicprocesses involved in cancers (adipocytokine and insulin signaling pathways, ubiquitin mediated proteolysis, aldosteroneregulated sodium reabsorption), hormones and proteins related to cancers (thyroid hormone, gonadotropin-releasinghormone (GnRH), chemokine signaling pathway, proteoglycans in cancer) and some highly attractive functions directlyassociated with EMT (TGF-β, tight junctions, gap junctions,adherence junctions, etc.).MiRNA profiles of exosomes resemble those of their parent cells. However, the top 10 different miRNAs of E- andM-exosomes were almost different with the top 10 differentmiRNAs of E- and M-cells (Table 1). Some miRNAs likemiR-5787, miR-4532 and miR-4488 were only selectivelypackaged into exosomes (Table 2) or up-regulated inM-exosomes when compared with M-cells (Table 1). Intriguingly, these miRNAs may target genes involved in EMTand cancer progression (Fig. 3d).E-exo vs Con-exomiRNA IDE-exoCon-exo up/down log2P-value(fold change) Exosomes derived from M-phenotype cancer cells promotethe transformation of recipient cells from E-phenotype toM-phenotypeThe above results showed that differentially expressedmiRNAs contained in exosomes may target genes relatedto EMT and the progression of cancer. Therefore, thefunction of exosomes coming from different cell types

Tang et al. BMC Genomics(2018) 19:802Page 8 of 14Table 2 miRNAs only detected in M-exosomes in pairedcomparisonsM-exo vs E-exomiRNA 0002.82E-06miR-487b-3p0.017447M-exo vs Con-exomiRNA 4795M-exo vs M-cellmiRNA 3 42was verified by a co-culture experiment. First, lipid staining and immunofluorescence microscopy were used toconfirm that exosomes derived from A549 cells canenter the same recipient tumor cells (A549) (Fig. 4a).Then we found the morphology of epithelial A549 cellsturned from being round into a spindle-like mesenchymal phenotype and lost intercellular junctions aftertreatment with M-exosomes (Fig. 4b).We next examinedthe effect of E- and M-exosomes on migration and invasion of E-phenotype A549 cells using a transwell system.When compared with the PBS –treated group, althoughboth E- and M-exosomes enhanced the migration and invasion of A549 cells, the effect of M-exosomes was more noticeable than that of E-exosomes (Fig. 4c, d). Additionally,the increased capacity of migration and invasion, EMT ischaracterized by a variation in EMT-related markers.Regarding the epithelial marker E-cadherin, although nosignificant change was found at the protein level, themRNA level decreased in A549 cells when they wereco-cultured with both E- and M-exosomes. However, protein and mRNA expression of N-cadherin and vimentinwere significantly up-regulated in A549 cells whenco-cultured not with E-exosomes but with M-exosomes(Fig. 4e, f). Similarly, up-regulated expression of vimentinbecame more pronounced after treatment with a higherconcentration of M-exosomes (Fig. 4f).To determine the effect of E/M-exosomes in other lungcancer cell lines, H1299 with higher malignant behaviorthan A549 was selected for functional analysis. In the intaketest, exosomes from 16HBE, A549, H1299 can be taken byH1299 cells (Fig. 5a). Morphologically, H1299 cells turnedfrom being round into a spindle-like mesenchymal phenotype after stimulation by M1/E2/M2-exosomes (Fig. 5b). Inthe transwell system, more migrated H1299 cells werefound in M1 and M2-exosome treated groups (Fig. 5c), andmore invaded cells were observed in E1, M1 andM2-exosomes (Fig. 5d) than PBS and Con-exosomestreated groups. M-exosomes showed higher ability to induce invasion and migration than E-exosomes for bothA549 and H1299. In addition to the expression of EMTmarkers, the protein (Fig. 5e) and mRNA levels (Fig. 5f) ofN-cadherin and vimentin of M1-exosomes group werehigher than PBS, Con and E1-exosomes groups. ThemRNA levels of N-cadherin (both 50 and 100 μg/ml concentration) and vimentin (only 100μg/ml) of theM2-exosome group were higher than those for the PBS,Con and E2-exosomes groups (Fig. 5f).Given that exosomes derived from A549 and H1299 cellscould also be taken up by 16HBE cells (Fig. 6a),EMT-related function and markers of 16HBE cells weremeasured after co-culturing with E/M-exosomes to investigate whether the M-exosomes could induce EMT ofnon-tumorigenic respiratory epithelial cells. Consistent withthe results for cancer cells, M-exosomes showed greater potential to enhance the mesenchymal morphological changes(Fig. 6b) and the invasion and migration abilities of 16HBEcells (Fig. 6c, d) than E- and Con-exosomes. The 16HBEcells treated with E1/M1/E2/M2-exosomes also showed increased protein expression of the mesenchymal marker,N-cadherin, when compared with PBS/Con-exosomestreated cells (Fig. 6e). Down-regulated expression ofE-cadherin became more pronounced after co-culture withE2 and M2-exosomes than other groups (Fig. 6e). Interestingly, the mRNA levels of vimentin and snail were moremarkedly elevated in 16HBE cells after E/M-exosomes treatment compared with that in A549 as acceptor cells (Fig. 6f).DiscussionExosomes serve as molecular messengers

ized on the Bio-Rad ChemiDoc XRS Imager system (Bio-Rad Laboratories, California, USA). Tang et al. BMC Genomics (2018) 19:802 Page 2 of 14. Wound healing assays Cells were wounded using a 200-μl sterile pipette tip. Subsequently, the cells were washed twice with PBS and