Lung CSC‐derived exosomal miR‐210‐3p contributes to a pro‐metastatic phenotype in lung cancer by targeting FGFRL1

Abstract Lung cancer has the highest mortality rate among human cancers, and the majority of deaths can be attributed to metastatic spread. Lung cancer stem cells (CSCs) are a component of the tumour microenvironment that contributes to this process. Exosomes are small membrane vesicles secreted by all types of cells that mediate cell interactions, including cancer metastasis. Here, we show that lung CSC‐derived exosomes promote the migration and invasion of lung cancer cells, up‐regulate expression levels of N‐cadherin, vimentin, MMP‐9 and MMP‐1, and down‐regulate E‐cadherin expression. Moreover, we verified that these exosomes contribute to a pro‐metastatic phenotype in lung cancer cells via miR‐210‐3p transfer. The results of bioinformatics analysis and dual‐luciferase reporter assays further indicated that miR‐210‐3p may bind to fibroblast growth factor receptor‐like 1 (FGFRL1); silencing FGFRL1 enhanced the metastatic ability of lung cancer cells, whereas overexpressing FGFRL1 suppressed metastasis. Taken together, our results provide new insights into a potential molecular mechanism whereby lung CSC‐derived exosomal miR‐210‐3p targets FGFRL1 to promote lung cancer metastasis. FGFRL1 may be a promising therapeutic target in lung cancer.

metastasis. Song et al 6 found that miR-26a-5p promoted lung cancer metastasis. Jiang et al 7 demonstrated that miR-19a-3p potentiates hepatocellular carcinoma metastasis. Studies of breast cancer and cervical cancer have shown that microRNAs may foster metastasis by enhancing migratory and invasive abilities and altering the expression of EMT-associated proteins such as E-cadherin, N-cadherin and vimentin. 8,9 Exosomes are extracellular vesicles (diameter 30-150 nm) that are secreted by all living cells. 10 They express specific surface markers, such as CD63, CD81 and Tsg101, and contain several types of bioactive materials, including microRNA, protein and DNA. 11 Recent reports suggest that tumour microenvironment-derived cells such as cancer-associated fibroblasts (CAFs), myeloid-derived suppressor cells (MDSCs) and mast cells may communicate with cancer cells and enhance their metastatic ability through the transfer of special mi-croRNA encapsulated in exosomes. [12][13][14] The metastatic potential of cancer cells may therefore be regulated by other types of cells in the tumour microenvironment through the transfer of exosomes carrying pro-metastatic microRNAs.
Cancer stem cells (CSCs) are a small subset of heterogeneous cells found in tumour tissue or cultured cell lines. Previous studies have suggested that lung CSCs may be isolated and enriched from surgically resected tumour tissue or lung cancer cell lines such as A549, PC-9 and H460. 15,16 These cells display obvious morphological differences from their parental lung cancer cells and express a stemness phenotype that includes enhanced ability for self-renewal, resistance to chemo/radiotherapy, high expression of stemness-associated markers such as CD133, ALDH1, Sox2 and Nanog, and increased metastatic potential. Due to these specific features, lung CSCs were considered to be the main mediators of lung cancer metastasis and recurrence. 17 One recent study of renal carcinoma showed that CSC-derived exosomes promoted EMT in renal carcinoma cells by transferring miR-19b-3p. 18 As one of the most important and essential components of the tumour microenvironment, whether lung CSC-derived exosomes carrying special microRNA involves in the pro-metastatic phenotype of lung cancer cells and its underlying molecular mechanism remains unclear.
In the present study, we identified and enriched lung CSCs from parental A549 cells and explored the functional role of lung CSCderived exosomes in regulation of the pro-metastatic phenotype of lung cancer cells. We also investigated the underlying molecular mechanism.

| Cell culture
Lung cancer cell lines including A549, NCI-H1703 and human normal lung epithelial cell line BEAS-2B were obtained from the Shanghai Cell Bank of the Chinese Academy of Sciences (Shanghai, China).
To generate suspended and stem-like sphere-growing cells, A549 cells were dissociated into single cells and seeded onto 6-well ultra-low attachment plates (Corning) at a concentration of 1 × 10 4 cells/well. Cells were incubated in DMEM/F12 (Gibco) supplemented with recombinant human epidermal growth factor (rhEGF, 10 ng/mL; Sigma), basic fibroblast growth factor (bFGF, 10 ng/mL; Sigma) and insulin (4 U/I; Sigma) for 12 days. After primary tumour spheres reached approximately 50-100 μm/sphere, the spheres were dissociated with Accutase (Invitrogen). The single cells obtained were cultured for another 12 days until secondary spheres had formed. After 5 passages, the spheres were collected, and characterization experiments were performed.

| Cell transfection with miR-210-3p mimic, miR-210-3p inhibitor, FGFRL1 siRNA or pcDNA3.1-FGFRL1
MiR-210-3p mimic, miR-210-3p inhibitor and corresponding negative controls (miR-NC and miR-inhibitor NC, respectively) were purchased from GenePharma. A small interfering RNA (siRNA) targeting human FGFRL1 mRNA, its negative control siRNA and pcDNA3.1-FGFRL1 as well as pcDNA3.1 blank vector were also purchased from GenePharma. Lipofectamine 3000 reagent (Invitrogen) was used according to the manufacturer's protocol under serum-free conditions. After 6-12 h of transfection, the liquid was abandoned and replaced with fresh medium, and cells were incubated for another 24 or 48 h. Cells were then collected, and transfection efficiency was verified by quantitative real-time polymerase chain reaction (qPCR) analysis or Western blot.

| Tumour sphere formation
Single cells derived from tumour spheres or parental A549 cells were seeded in 96-well ultra-low plates at a density of 50 or 100 cells per well, respectively. After 12 days of non-adhesive culture, the number of spheres with >50 µm/sphere was counted, and representative morphology was captured under light microscopy (Leica). Sphereforming efficiency (SFE) was calculated as the number of spheres formed divided by the initial number of single cells plated and expressed as a percentage.

| Quantitative real-time PCR
Total RNA from cells or exosomes was extracted using TRIzol reagent (Invitrogen). The primer sequences for ALDH1, Sox2, Nanog, CD133, BDNF, E2F3, FGFRL1, KCMF1, NDUFA4, GAPDH, miR-210-3p and U6 were designed and synthesized by GenePharma. The primer sequences are listed in Table 1. RNAs were reverse-transcribed with PrimeScript Reverse Transcriptase (Takara) or microRNA first-strand cDNA synthesis (Sangon Biotech); qPCR analysis was performed with the SYBR Green Detection Kit (Takara) or the MicroRNA qPCR Kit (SYBR Green Method) (Sangon Biotech). All processes were performed according to the manufacturer's instructions. GAPDH and U6 were used as endogenous controls.

| Immunofluorescent staining assay
Tumour spheres were collected and fixed with 4% paraformaldehyde for 10 min before they were embedded in low-melting point agarose for 30 min at 4°C. The coagulated agarose was embedded in OCT compound (Tissue Tek). Frozen blocks were cut into section of 5-µm thickness. A549 cells were plated onto Matrigel-coated glass coverslips before fixation with 4% paraformaldehyde for 10 min at room temperature, followed by washes in phosphate-buffered saline (PBS). All cells were permeabilized with 0.5% Triton X-100 for 15 min at room temperature and blocked in 5% BSA for 30 min before incubation with CD133 and ALDH1 (1:300, Beijing Biosynthesis Biotechnology Co., LTD). Sections were incubated in primary antibodies overnight at 4℃ and then incubated with fluorescencetagged mouse or rabbit secondary antibodies. DAPI (Sigma-Aldrich) was used for nuclear staining. Fluorescence images were visualized with a fluorescence microscope (Leica).

| Western blot analysis
Cells or exosomes were treated with RIPA lysis buffer supplemented with protease inhibitor cocktail (Beyotime Biotechnology). Samples containing 20 μg protein were separated and transferred to PVDF membranes, and then blocked in 5% non-fat milk at room temperature for 1 h. The PVDF membranes were incubated with primary antibodies for rabbit anti-CD133, anti-Nanog, anti-Sox2, anti- Co., LTD), which was used at a 1:5000 dilution. Protein levels were detected with chemiluminescent HRP substrate and a Western blot analysis system (Universal Hood II, Bio-Rad).

| Exosome isolation and quantification
For lung cancer cell culture, the medium was changed to RMPI-1640 or DMEM containing 10% exosome-free FBS (System Biosciences) for 48 h, until cells reached 90% confluency and were subsequently collected. For tumour spheres, medium was collected from supernatant with approximately 2 × 10 6 cells/mL. After using a 0.22-μm filter (Millipore) to remove cellular debris, the medium obtained was ultra centrifuged at 120 000 × g, at 4°C for 60 min, to pellet the Nanog

| Transmission electron microscopy
The purified exosomes were prepared by mixing with an equal volume of 4% paraformaldehyde and deposited on Formvar carboncoated copper grids. Grids were stained with 2% uranyl acetate for 15 min, air-dried and then observed by transmission electron microscopy (TEM; FEI Tecnai 20; Philips).

| Nanoparticle tracking analysis
To measure the size of the purified exosomes, they were resuspended in PBS. NanoSight NS300 (Malvern Instruments) nanoparticle tracking analysis (NTA) software was then used to visualize and measure particle size.

| PKH26 labelling of exosomes
To observe the interaction between exosomes and lung cancer cells, exosomes were labelled with PKH26 (a membrane fluorescence dye that is red in colour) (Sigma-Aldrich). After co-incubation with PKH26-labelled exosomes for 6 h, lung cancer cells were counterstained with DAPI. Staining patterns were visualized with an inverted fluorescence microscope linked to a camera (Leica).

| Detection of miR-210-3p transfer
Tumour spheres were transiently transfected with fluorescein amidite (FAM)-labelled miR-210-3p mimic (GenePharma) with Lipofectamine 3000. After transfection for 48 h, the culture medium was collected F I G U R E 1 Identification of stemness phenotype in tumour spheres derived from parental A549 cells. A, The diagram shows the SFE of tumour sphere-derived cells and parental A549 cells, respectively. B, Representative images of sphere formation from tumour spherederived cells and parental A549 cells cultured in non-adhesive conditions for 12 days. Scale bars: 100 µm. C, Tumorigenesis of tumour sphere-derived cells or parental A549 cells in nude mice (n = 3). D, qPCR analysis of ALDH1, CD133, Nanog and Sox2. E, Western blot analysis of ALDH1, CD133, Nanog and Sox2. F, Immunofluorescent staining of tumour spheres and parental A549 cells.

| MTT assay
Lung cancer cells were seeded in 100 μL of medium/well (2 × 10 3 cells per well) in 96-well plates. After co-incubation with various con-

| Cell migration assay
A wound-healing assay was performed to assess migratory ability under various conditions. Briefly, cells were seeded in 6-well plates to create a confluent monolayer. A scratch wound was made with a sterile pipette tip, after which plates were incubated for 24 or 48 h with or without exosomes. Images were captured under a camera equipped with a light microscope (Leica).

| Transwell assay
Cell migration and invasion were evaluated using Transwell cham-

| Statistical analysis
Differences were examined using Student's t test or one-way analysis of variance. The data are presented as means ± SD.
Values of P < .05 were considered statistically significant. All statistical analyses were performed using SPSS version 17.0 software (IBM).

| Identifying stemness phenotype in A549 cellderived tumour spheres
Self-renewal ability is the most important characteristics of CSCs.
As shown in Figure 1A, through limiting dilution analysis, we demonstrated that the SFE of tumour sphere-derived cells was significantly increased, compared to that of parental A549 cells. Indeed, single cells derived from tumour spheres were able to form new spheres, whereas parental A549 cells had no such effect. Most A549 cells present in the non-adhesive culture appeared to be dead ( Figure 1B). When we evaluated tumorigenesis in vivo, we found that inoculation with 10 2 , 10 3 , 10 4 or 10 5 tumour spheres generated tumours in 1/3, 2/3, 3/3 and 3/3 of inoculations, respectively.
Stemness-associated markers including ALDH1, Sox2, Nanog and CD133 were closely involved with maintenance of the stemness phenotype in lung CSCs. Through qPCR and Western blot analysis, we verified that the expression levels of these markers were significantly higher in tumour spheres than in parental A549 cells ( Figure 1D-E). Immunofluorescent staining analysis further revealed that the intensity of ALDH1 and CD133 fluorescence was higher in tumour spheres, compared to parental A549 cells ( Figure 1F). CSCs acquire enhanced metastatic ability during the EMT. The results of the Transwell assays performed in this study showed that the number of migrated or invaded cells derived from tumour spheres was greatly increased, compared to the number of migrated or invaded cells derived from parental A549 cells ( Figure 1G-H). Western blot F I G U R E 3 Exosomes exert pro-metastatic effects in lung cancer cells. A, The diagram shows the migratory ability of A549 and NCI-H1703 cells after treatment with lung CSC-derived exosomes. B, Transwell invasion of A549 and NCI-H1703 cells after treatment with lung CSC-derived exosomes. Scale bars: 100 μm. C, Western blot analysis for E-cadherin, N-cadherin, vimentin, MMP-9 and MMP-1. Control vs. 40 or 80 μg/mL exosomes, respectively. *P < .05. **P < .01. Data are mean ± SD from three independent experiments performed in triplicate analysis of tumour spheres further showed that the expression of EMT-associated proteins including N-cadherin, vimentin, MMP-9 and MMP-1 was up-regulated, whereas E-cadherin expression was down-regulated ( Figure 1I). These results suggest that tumour spheres derived from parental A549 cells expressed the stemness phenotype. These cells appeared to have dedifferentiated into lung CSCs.

| Lung cancer cells internalized lung CSCderived exosomes
To investigate the effects of exosomes in lung cancer cells, we purified exosomes from lung CSCs. As shown in Figure 2, the characteristics of exosomes were verified through TEM, NTA and Western blot analysis. TEM observation revealed that most vesicles had a typical cup-shaped morphology (Figure 2A). NTA analysis showed that the average diameter of exosomes ranged from 30 to 150 nm ( Figure 2B). Western blot analysis further demonstrated that these exosomes expressed special markers including CD63, TSG101 and CD81 ( Figure 2C). Next, we labelled these exosomes with PKH26 to observe their interactions with lung cancer cells. As shown in Figure 2D, lung cancer cells internalized lung CSC-derived exosomes, indicating that CSC-derived exosomes may regulate lung cancer cells. Interestingly, the results of MTT analysis showed that when lung cancer cells were co-incubated with lung CSCderived exosomes for more than 2 days, cell proliferation increased ( Figure 2E).

| Lung CSC-derived exosomes contribute to a pro-metastatic phenotype in lung cancer cells
To clarify the possibility of a pro-metastatic role for exosomes, lung cancer cells were co-incubated with lung CSC-derived exosomes.
Our results showed that these exosomes enhanced the migration and invasion of lung cancer cells in a dose-dependent manner ( Figure 3A,B and Figure S1). Moreover, through Western blot analysis of lung cancer cells treated with lung CSC-derived exosomes, we demonstrated that expression levels of N-cadherin, vimentin, MMP-9 and MMP-1 were up-regulated, whereas E-cadherin expression was down-regulated ( Figure 3C).

| Exosomal miR-210-3p regulated the metastatic potential of lung cancer cells
Recent reports indicate that miR-210-3p plays a central role in the metastasis of prostate cancer and renal cell carcinoma. 19,20 More importantly, mesenchymal stem cell-derived exosomes promoted lung cancer metastasis via the transfer of miR-210-3p. 21 In this study, we demonstrated that lung CSC-derived exosomes promote lung cancer metastasis. Based on the results obtained, we speculate that exosomes may be associated with the transfer of miR-210-3p. We therefore analysed miR-210-3p levels in cells and their derived exosomes. As shown in Figure 4A, miR-210-3p levels were significantly higher in lung CSCs, compared with lung cancer cells.

| FGFRL1 may be a target of miR-210-3p in lung cancer cells
To further verify the molecular mechanism by which miR-210-3p promotes a pro-metastatic phenotype in lung cancer cells, three bioinformatics software programs (TargetScan, miRTarBase and miDB) were used to identify possible potential downstream targets of miR-210-3p. As shown in Figure 6A, five potential genes (BDNF,

| D ISCUSS I ON
During recent years, new therapeutic agents have emerged for lung cancer patients. Updated therapeutic strategies include the use of third-generation EGFR/TKI or PD-1/PD-L1 checkpoint inhibitors.
However, the overall survival of patients with advanced lung cancer remains low, and the clinical prognosis is poor. 22 Metastasis is the most common cause of death in this patient population. The suppression of lung cancer metastasis may be a promising and feasible therapeutic strategy for improving survival among lung cancer patients.
Lung CSCs play an important role in tumour metastasis and recurrence. In this study, the tumour spheres derived from parental A549 cells overexpressed stemness-associated markers including CD133, Nanog, Sox2 and ALDH1, which are associated with the lung CSC phenotype. 23,24 The gold standard for the differentiation of CSCs from parental cancer cells is the capacity for self-renewal. 25 By analysing SFE in vitro, we verified that SFE was significantly increased in tumour spheres, compared to parental A549 cells.
Another characteristic associated with the capacity for self-renewal is the enhanced tumorigenesis of lung CSCs in vivo. One previous study of a mouse model indicated that inoculation with only 10 2 lung CSCs resulted in xenograft formation. 26 In this study, we demonstrated that inoculation with 10 2 tumour spheres generated xenografts, whereas there was no xenograft formation in mice inoculated subcutaneously with 10 5 parental A549 cells. These findings provide further support for the potent self-renewal of tumour spheres. EMT has been confirmed to be an essential step for metastasis. E-cadherin, N-cadherin and vimentin have been identified as the main proteins associated with cell migration. 27 MMP-9 and MMP-1 are proteolytic enzymes that facilitate cell invasion by degrading extracellular matrix, which leads to metastasis. 28 Previous reports indicated that lung CSCs were endowed with EMT characteristics and increased risk for metastasis. 29 Indeed, in our study, we found that migratory and invasive abilities were increased in tu- MDSCs releasing exosomal miR-126a were found to promote lung F I G U R E 5 Exosomal miR-210-3p regulates the metastatic potential of lung cancer cells derived from lung CSCs transfected with miR-210-3p mimic. A, qPCR analysis of miR-210-3p levels in lung CSCs and their derived exosomes after transfection of lung CSCs with miR-210-3p mimic (miR-NC vs. miR-210-3p mimic). B, qPCR analysis of miR-210-3p levels in A549 and NCI-H1703 cells after co-incubation with exosomes (80 μg/mL) derived from lung CSCs transfected with miR-210-3p mimic or miR-NC. C, The diagram shows the migratory ability of A549 and NCI-H1703 cells after co-incubation with exosomes (80 μg/mL) derived from lung CSCs transfected with miR-210-3p mimic or miR-NC. D, Transwell invasion of A549 and NCI-H1703 cells after co-incubation with exosomes (80 μg/mL) derived from lung CSCs transfected with miR-210-3p mimic or miR-NC.   and MMP-1, with down-regulated E-cadherin expression.
Our investigation of the molecular mechanism by which lung CSC-derived exosomal miR-210-3p regulates lung cancer metastasis revealed that FGFRL1 may be the functional target of miR-210-3p. In this study, we demonstrated that lung CSC-derived exosomes markedly. In combination, these results suggest that FGFRL1 acts in a tissue-and cell-specific manner. In lung cancer, FGFRL1 may act as a tumour suppressor, but further investigation is needed.

| CON CLUS ION
To sum up, our results preliminarily demonstrate that lung CSCs contribute to a pro-metastatic phenotype in lung cancer cells by transferring exosomes carrying miR-210-3p. FGFRL1 may be a target of miR-210-3p in the regulation of lung cancer metastasis.

ACK N OWLED G EM ENTS
The present study was supported by the National Natural Science

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Additional data and materials may be requested from the corresponding author on reasonable request.