Transfer of Extracellular Vesicle-Associated-RNAs Induces Drug Resistance in ALK-Translocated Lung Adenocarcinoma

Anaplastic lymphoma kinase (ALK) translocation is an actionable mutation in lung adenocarcinoma. Nonetheless tumour consists of heterogeneous cell subpopulations with diverse phenotypes and genotypes, and cancer cells can actively release extracellular vesicles (EVs) to modulate the phenotype of other cells in the tumour microenvironment. We hypothesized that EVs derived from a drug-resistant subpopulation of cells could induce drug resistance in recipient cells. We have established ALK-translocated lung adenocarcinoma cell lines and subclones. The subclones have been characterized and the expression of EV-RNAs determined by quantitative polymerase chain reaction. The effects of EV transfer on drug resistance were examined in vitro. Serum EV-RNA was assayed serially in two patients prescribed ALK-tyrosine kinase inhibitor (ALK-TKI) treatment. We demonstrated that the EVs from an ALK-TKI-resistant subclone could induce drug resistance in the originally sensitive subclone. EV-RNA profiling revealed that miRNAs miR-21-5p and miR-486-3p, and lncRNAs MEG3 and XIST were differentially expressed in the EVs secreted by the resistant subclones. These circulating EV-RNA levels have been found to correlate with disease progression of EML4-ALK-translocated lung adenocarcinoma in patients prescribed ALK-TKI treatment. The results from this study suggest that EVs released by a drug-resistant subpopulation can induce drug resistance in other subpopulations and may sustain intratumoural heterogeneity.


Introduction
Rearrangement of chromosome 2 leads to fusion of the echinoderm microtubule-associated protein-like 4 (EML4) and anaplastic lymphoma kinase (ALK) gene to form the constitutively active EML4-ALK oncoprotein. Although EML4-ALK fusion occurs in only 3-5% of lung adenocarcinomas [1], it is a therapeutic target in non-small cell lung cancer (NSCLC).
The first-generation ALK-tyrosine kinase inhibitor (TKI) crizotinib is a small molecule TKI of ALK, MET and ROS1 kinases. Crizotinib was previously the recommended first-line treatment for ALK-translocated NSCLC. Although most cases of ALK-translocated NSCLC respond initially to crizotinib, disease progression occurs approximately one year after treatment. The resistance   Table 1. IC 50 values of different FA34 and FA121 -parental and -resistant subclones against the three ALK-TKIs tested. (A) The FA34 and FA121 subclones were incubated with crizotinib, ceritinib or alectinib (1 nM to 100 µM) for 72 h. Cell viability was determined by 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) assays. The (B) crizotinib or (C) ceritinib resistant subclones of FA34 and FA121 were treated with crizotinib, ceritinib or alectinib (1 nM to 100 µM) for 72 h. Cell viability was determined by MTT assay. The values in brackets indicate the fold-changes in IC 50 compared with the respective subclones before prolonged TKI exposure. For secondary mutations, the ALK kinase domain was amplified and was sequenced to detect secondary mutations. Presence of ALK amplifications in these resistant subclones was determined by qRT-PCR. Wild-type (WT). All subclones of the two ALK cell lines (FA34 and FA121) were continuously incubated with crizotinib (Table 1B) or ceritinib (Table 1C) with either stepwise escalation of drug concentrations or high bolus concentrations. For FA34 subclones ( Figure 1F), only FA34.3 and FA34.5 survived following continuous incubation with crizotinib. The IC 50 of FA34.3 against crizotinib increased from 0.4 µM to approximately 20 µM (a 50-fold increase), while the IC 50 of FA34.5 increased from 2 µM to 20 µM (a 10-fold increase) in the resistant subclones (named FA34.3SCr, FA34.3HCr, FA34.5SCr and FA34.5HCr, respectively).
All four subclones of FA121 ( Figure 1G) survived after prolonged incubation with crizotinib, and their IC 50 against crizotinib increased from 2-fold to 130-fold. Continuous ceritinib exposure eliminated most of the FA34 subclones but not those of FA34.3 and FA34.5 (named FA34.3SCe, FA34.3HCe, FA34.5SCe and FA34HCe, respectively). The IC 50 of resistant FA34.5 increased from 0.4 µM to 5 µM or 2 µM in two treatment methods. All the FA121 subclones survived after prolonged crizotinib or ceritinib treatment. These results also suggested that different levels of intratumoural heterogeneity among patients may be associated with different clinical outcomes during ALK-TKI treatment.
Direct sequencing of the cDNA amplicon of the ALK kinase domain revealed a resistant mutation C1156S in FA121.3HCr; IC 50 of the recipient cells against crizotinib increased from 0.1 µM to approximately 13 µM (Table 1B). Another resistant mutation T1151M was found in several ceritinib-treated cell lines including FA121.3SCe and FA121.4SCe, and also in FA121.4HCe (Table 1C). ALK gene copy number gain (CNG) representing ALK gene amplifications was found in all FA34-resistant subclones regardless of treatment with crizotinib or ceritinib. Some FA121-resistant subclones also acquired ALK-CNG after treatment with crizotinib or ceritinib.

Identification of ALK-TKI Resistance-Related EV-RNAs in Lung Adenocarcinoma Subclones
Next we compared the intracellular ( Figure 2A) and EV-associated ( Figure 2B) expression of miRNAs in the ALK-TKI-sensitive subclones FA34.8 and the drug-resistant subclones FA34.5. This would allow for identification of EV-miRNAs that were differentially expressed in drug-resistant cells and in the EVs they released.

Characterization of Isolated EVs
EVs were isolated from conditioned medium (CM) of FA34 or from serum samples from patients with lung cancer and prescribed ALK-TKI treatment. The average size of EVs from CM and serum was 109 nm and 84 nm respectively ( Figure 2D). The purity of EVs was verified by the presence of EV protein marker CD63, and the absence of non-EV protein marker GP96 and GAPDH ( Figure 2E). To confirm EV uptake, FA34.8 was incubated with isolated EVs labelled with DiD for one hour; exosome uptake in recipient cells was demonstrated by fluorescence microscopy ( Figure 2F). The cells pre-treated with EIPA, a Na + -H + exchange inhibitor that blocks micropinocytosis, demonstrated markedly reduced exosome uptake ( Figure 2F, lower left). was 109 nm and 84 nm respectively ( Figure 2D). The purity of EVs was verified by the presence of EV protein marker CD63, and the absence of non-EV protein marker GP96 and GAPDH ( Figure 2E). To confirm EV uptake, FA34.8 was incubated with isolated EVs labelled with DiD for one hour; exosome uptake in recipient cells was demonstrated by fluorescence microscopy ( Figure 2F). The cells pre-treated with EIPA, a Na + -H + exchange inhibitor that blocks micropinocytosis, demonstrated markedly reduced exosome uptake ( Figure 2F, lower left).

Transfer of EV-RNAs Induced Drug-Resistance and Cell Migration
To evaluate the functional effects of EV transfer between TKI-sensitive and -resistant subclones, EVs were isolated from the TKI-resistant subclones FA34.5. These EVs were transferred to TKI-sensitive subclone FA34.8. The intracellular level of miR-21-5p was decreased while miR-486-3p in the recipient cells was increased when compared with cells treated with EVs from the same subclone FA34.8 ( Figure 3A, upper panel). The expression of PTEN and MSH2 was down-regulated in EV-recipient cells ( Figure 3A, lower panel). Nonetheless resistant cells produced EVs containing a relatively low expression level of miR-21-5p that could not lead to down-regulation of intracellular level of miR-21-5p in the recipient cells. Other factors that could actively reduce the miR-21-5p, and subsequently PTEN and MSH2 in recipient cells must be present in EVs released from the resistant cells. Recent studies suggest that lncRNAs act as competitive endogenous RNA (ceRNA) of some miRNAs [17]. Two lncRNAs, MEG3 and XIST have been proposed to bind to miR-21-5p [18,19]. We therefore determined the expression of two lncRNAs MEG3 and XIST in the EVs released by different FA34 subclones. Both lncRNAs were highly expressed in the EVs of resistant subclones compared with those from the sensitive subclones ( Figure 3B) indicating that these lncRNAs may be incorporated into EVs produced from resistant cells and transferred to sensitive cells, thus competitively binding with miR-21-5p in the recipient cells. To evaluate the functional effects of EVs transfer between subclones, the TKI-sensitive subclones FA34.8 or FA121.1 were exposed to EVs isolated from TKI-resistant subclones. Transfer of EVs produced from resistant subclones significantly increased drug resistance of sensitive subclones ( Figure 3C, upper panel). To determine whether this transfer of resistance phenotype was subclone specific, we examined the effects of allogeneic EV transfer between the two ALK-translocated cell lines ( Figure 3C, lower panel). We transferred the EVs from FA34-resistant subclone to FA121-sensitive subclone, and vice versa. The resistance-enhancing effects of these EVs were even transferrable between different cell lines. In addition, a marginal increase in cell migration was observed after TKI-sensitive cells were treated with EVs from resistant cells (Supplementary Figure S1).

Transfer of EV-RNAs Induced Drug-Resistance and Cell Migration
To evaluate the functional effects of EV transfer between TKI-sensitive and -resistant subclones, EVs were isolated from the TKI-resistant subclones FA34.5. These EVs were transferred to TKIsensitive subclone FA34.8. The intracellular level of miR-21-5p was decreased while miR-486-3p in the recipient cells was increased when compared with cells treated with EVs from the same subclone FA34.8 ( Figure 3A, upper panel). The expression of PTEN and MSH2 was down-regulated in EVrecipient cells ( Figure 3A, lower panel). Nonetheless resistant cells produced EVs containing a relatively low expression level of miR-21-5p that could not lead to down-regulation of intracellular level of miR-21-5p in the recipient cells. Other factors that could actively reduce the miR-21-5p, and subsequently PTEN and MSH2 in recipient cells must be present in EVs released from the resistant cells. Recent studies suggest that lncRNAs act as competitive endogenous RNA (ceRNA) of some miRNAs [17]. Two lncRNAs, MEG3 and XIST have been proposed to bind to miR-21-5p [18,19]. We therefore determined the expression of two lncRNAs MEG3 and XIST in the EVs released by different FA34 subclones. Both lncRNAs were highly expressed in the EVs of resistant subclones compared with those from the sensitive subclones ( Figure 3B) indicating that these lncRNAs may be incorporated into EVs produced from resistant cells and transferred to sensitive cells, thus competitively binding with miR-21-5p in the recipient cells. To evaluate the functional effects of EVs transfer between subclones, the TKI-sensitive subclones FA34.8 or FA121.1 were exposed to EVs isolated from TKI-resistant subclones. Transfer of EVs produced from resistant subclones significantly increased drug resistance of sensitive subclones ( Figure 3C, upper panel). To determine whether this transfer of resistance phenotype was subclone specific, we examined the effects of allogeneic EV transfer between the two ALK-translocated cell lines ( Figure 3C, lower panel). We transferred the EVs from FA34-resistant subclone to FA121-sensitive subclone, and vice versa. The resistance-enhancing effects of these EVs were even transferrable between different cell lines. In addition, a marginal increase in cell migration was observed after TKI-sensitive cells were treated with EVs from resistant cells (Supplementary figure S1).

Prognostic Value of Circulating EV-RNAs in ALK-Translocated Lung Adenocarcinoma Patients
The expression level of circulating EV-RNAs in two ALK-translocated lung adenocarcinoma patients at different stages of treatment ( Figure 4A,B) were assayed (complete profile in Figure  S2A,B). The expression patterns of EV-RNAs were very similar in both patients. Serum EV-miR-21-5p showed a steady decrease during the course of treatment, through baseline, stable disease and disease progression; while EV-miR-486-3p, lncRNA MEG3 and XIST increased throughout the course of treatment. These results suggest that these serum EV-RNAs may serve as circulating biomarkers to enable early detection of disease recurrence in patients with ALK-translocated lung adenocarcinoma.
patients at different stages of treatment ( Figure 4A and B) were assayed (complete profile in Figure  S2A,B). The expression patterns of EV-RNAs were very similar in both patients. Serum EV-miR-21-5p showed a steady decrease during the course of treatment, through baseline, stable disease and disease progression; while EV-miR-486-3p, lncRNA MEG3 and XIST increased throughout the course of treatment. These results suggest that these serum EV-RNAs may serve as circulating biomarkers to enable early detection of disease recurrence in patients with ALK-translocated lung adenocarcinoma.

Discussion
The results of this study suggest that transfer of EV-RNAs can sustain tumour subclonal heterogeneity. Although ALK inhibitors are effective in most patients with EML4-ALK positive NSCLC, the eventual development of acquired resistance remains a major concern in the treatment of EML4-ALK translocated lung adenocarcinomas. Lung tumours consist of different subpopulations of cancer cells with different phenotypic expression of cancerous properties as well as drug sensitivity [20]. Under selective pressure from chemotherapy or targeted-therapy, the interactions between subclones of cancer cells may affect treatment response and efficacy. In the lung adenocarcinoma cell line models established in this study, different FA34 subclones showed variable sensitivity to ALK-TKI, and two subclones of FA34 were able to develop resistance to ALK-TKIs. The first round of drug treatment may have introduced a selective pressure and could have suppressed the TKI-sensitive subclones (i.e., FA34.4, .8, .11, .12, .13 and .14). The surviving resistant subclones (i.e., FA34.3 and .5) could become the dominant clone after treatment. Nonetheless the results of in vitro EV transfer experiments suggest that the TKI-resistant subclones could release EV with low miR-21-5p and high miR-486-3p, and a high level of lncRNAs MEG3 and XIST. The EV-RNAs then modulated TKI-sensitivity by inhibition of tumour suppressors PTEN and MSH2 of the recipient subclones and enhanced their survival. This observation may in part explain the intratumoural heterogeneity of chemo-sensitivity.
In this study, we investigated the biological functions of EV-miRNAs and EV-lncRNAs. The role of miR-21-5p has been extensively studied in various cancers [21]. The fact that over-expression of miR-21-5p has been found in most tumours suggests that it may be an oncomiR that contributes to cancer cell proliferation, inhibition of apoptosis, promotion of angiogenesis [22] and metastasis [23]. The large-scale prospective clinical study BIOMILD (NCT02247453) selected 24 previously identified plasma miRNAs including miR-21 as diagnostic signatures for lung cancer screening. A recent study using next-generation sequencing (NGS) has shown that exosomal miR-21-5p is downregulated rather than upregulated in lung cancer patients compared with healthy controls. These conflicting results regarding the expression of miR-21-5p between tissue and EVs [11,24] suggests its distinct roles in lung cancer. In our study, the contradictory expression pattern of miR-21-5p (high intracellular expression but low in EV) in resistant cells led us to further explore other EV-associated components, including EV-lncRNAs, that could affect the level of miR-21-5p in EV-recipient cells. The resistant cells produced only EVs with relatively low expression of miR-21-5p and could not have led to down-regulation of intracellular miR-21-5p, and subsequently PTEN and MSH2 in the recipient cells. The regulatory functions of lncRNAs XIST and MEG3 mediated sponging of miR-21-5p adding another layer of complexity in subclonal communications via EVs, but also highlighted the precise control EV-cargo sorting.
For miR-486, both -3p and -5p strands have been proposed as oncomiR [25,26]. Intracellular miR-486-5p has been shown to directly suppress NF-κB-negative regulators and promote cancer aggressiveness through activation of NF-κB [25]. The prognostic power of exosomal miR-486-5p in NSCLC has also been demonstrated in a recent study using NGS [11]. Since a single protein can be regulated by multiple miRNAs, a cellular response such as drug resistance may involve numerous miRNAs. As cancer cells often secrete several times more EVs than normal cells [27], a panel of circulating EV-miRNA from NSCLC patients could have the potential for use as non-invasive circulating biomarkers for lung cancer patients [28].
Data from recent clinical trials show that ceritinib and alectinib, two second-generation ALK inhibitors, are superior to crizotinib in the first-line setting in terms of progression-free survival, especially for patients with brain metastases [29,30]. The clinical use of ALK-TKIs could be enhanced with companion biomarkers. Liquid biopsy, especially with blood sampling, may reveal the changes in oncogenic mutations along a treatment course. The resistant cell-derived EV-RNAs identified in this study may provide a source of biomarkers for monitoring of disease and treatment course. The use of serum EV-RNAs as promising biomarkers has emerged for various cancers because of their stability in circulation [10]. In contrast to free circulating miRNA, EV-miRNAs are more stable. miRNA could be packaged and enriched in EVs and then actively secreted in extracellular space for trans-cellular communication, taking part in oncogenic processes like tumour metastasis [23,31], angiogenesis [22], and chemoresistance [32]. Since specific RNA sorting [33] and cell-free miRNA biogenesis machinery [34] has been found in EVs, these mechanisms may indicate that EV-RNAs should better represent the cellular conditions than the randomly released cell-free circulating RNA [11,24]. There are some limitations in this study. First, the EV-RNA signatures, low EV-miR-21-5p and high miR-486-3p, MEG3 and XIST identified in this study require clinical validation in a larger patient cohort to confirm their prognostic relevance in ALK-translocated lung cancer. Second, the results from this study are relevant to ALK-translocated lung cancer. Whether the same EV-RNA profiles are evident in other histological or molecular subtypes of lung cancer warrants further investigation. It is worth mentioning that the two cell lines used in the present study were established from malignant pleural effusions of ALK-TKI-naïve patients previously treated with standard platinum-based chemotherapy. Undoubtedly the EV-cargo was affected by the chemotherapy through multiple mechanisms [35]. The extent to which this could impact the predictive power of the selected exosomal miRNAs should be validated in a larger cohort. Further experiments are also warranted to explore the biological effects of EVs on relevant cancer metabolomics and epigenomic regulation of tumour heterogeneity. The novel ALK-translocated lung cancer cell lines together with their subclones will enable further understanding of the development of drug resistance within the heterogeneous tumour microenvironment. This will shed light on drug resistance mechanisms in ALK-translocated lung cancer.

ALK-Translocated Lung Adenocarcinoma Cell Lines
All clinical specimens were collected after informed written consents had been given by the respective subjects. The study was approved by the University of Hong Kong/Hong Kong Hospital Authority Hong Kong West Cluster Institutional Review Board/Ethics Committee (UW 11-326).
The EML4-ALK lung adenocarcinoma cell line FA34 was established from a malignant pleural effusion of a 34 year-old female non-smoker with left lower lobe adenocarcinoma as described in a previous report [36]. Another EML4-ALK lung adenocarcinoma cell line FA121 was established from the pleural fluid of a 44-year-old male non-smoker with stage IV EML4-ALK translocated lung adenocarcinoma. Both patients had received standard platinum-based chemotherapy but had not been treated with an ALK-inhibitor prior to collection of pleural fluid.
The subclones of FA34 and FA121 were selected with single cell cloning either by serial dilution or by cloning cylinders. Eight subclones were generated from the FA34 parental cell line (FA34P) (the subclones were named FA34.3, FA34.4, FA34.5, FA34.8, FA34.11 (established by serial dilution), and FA34.12, FA34.13, and FA34.14 (established by cloning cylinders). Four subclones were generated from the FA121 parental cell line (FA121.P) (the subclones were named FA121.1, FA121.3, FA121.4, FA121.5 and they were all established by serial dilution methods). The parental and the subclones of FA34 were maintained in RPMI 1640 medium with 10% fetal bovine serum (FBS) (Gibco, Grand Island, NY, USA). The FA121 parental and subclones were maintained in ACL4 medium [37] supplemented with 10% FBS. All cell lines were cultured at 37 • C in a 5% CO 2 incubator. All cell lines were established with more than 50 passages as an adherent monolayer. The identity and purity of all parental cell lines and the subclones were authenticated by polymorphic short tandem repeat (STR) DNA profiling assay (Genetica DNA Laboratories Inc, Burlington, NC, USA).

Fluorescence In Situ Hybridization (FISH)
Cultured cells were harvested and fixed onto positively-charged glass slides. ALK FISH was performed with the Vysis ALK Dual Specific Break Apart FISH Probe Kit (Abbott Molecular, Des Plaines, IL, USA) according to the manufacturer's protocol. The images were analyzed and captured by the Isis FISH imaging system (MetaSystems, Altlussheim, Germany).

Reverse Transcription-Polymerase Chain Reaction (RT-PCR)
To identify the variation of EML4-ALK in these newly established lung adenocarcinoma cell lines and subclones, RT-PCR was performed with primers specific to EML4 intron 13 and ALK intron 19 [38]. Total RNA was extracted from cultured cells using Trizol (Invitrogen, Carlsbad, CA, USA).
Complementary DNA (cDNA) was synthesized using the QuantiTect reverse transcription kit (Qiagen, Hilden, Denmark) according to the manufacturer's instruction. The types of EML4-ALK fusion mRNA variants were identified by RT-PCR using the Fusion-RT-S and Fusion-RT-AS primers [38] (Table S1). PCR was performed with initial denaturation at 95 • C for 4 min, followed by 40 cycles of amplification (95 • C for 30 s, 55 • C for 30 s, and 72 • C for 2.5 min), and final extension at 72 • C for 10 min. The PCR products were resolved on 1.5% agarose gel and stained with SYBRsafe.

Establishment of Lung Adenocarcinoma Cell Lines with Resistance to Crizotinib or Ceritinib
To study the ALK-TKI resistant mechanisms, we established ALK-TKI-resistant cell lines. The subclones of FA34 and FA121 were incubated with crizotinib (Cr) or ceritinib (Ce) (Selleck Chemicals, Houston, TX, USA) with stepwise escalation (labelled with the suffix S) of drug concentrations (crizotinib: 0.02, 0.05, 0.2 µM; ceritinib: 0.01, 0.02, 0.04 µM), or high bolus concentrations (labelled with the suffix H) (crizotinib: 0.4 µM, ceritinib: 0.1 µM). IC50 values were determined by non-linear regression curve fitting using GraphPad Prism software 6.01. The IC 50 of the drug-resistant cell lines against different ALK-TKIs was determined after 20 passages. The cells that survived after prolonged exposure to TKIs were considered drug-resistant cell lines. Drug-resistant cell lines treated with first-generation ALK-TKI crizotinib were annotated with suffix Cr; resistant cell lines treated with second-generation ALK-TKI ceritinib were annotated with suffix Ce.

Detection of Secondary ALK Mutations or Amplifications
Secondary mutations in resistant cell lines were detected by direct sequencing of ALK kinase domain cDNA amplicon. In brief, the primers (EML4-E20-F and EML4-ALK-4193-R) (Table S1) flanking the ALK kinase domain coding region were used for PCR amplification and the amplicon size was 1055 bp. The resolved PCR product extracted from agarose gel was subjected to direct sequencing (Tech Dragon Ltd., Hong Kong, China) with the primer 3221-F (Table S1). ALK amplification was determined by quantitative PCR (qPCR) assay using QuantiFast SYBR Green PCR (Qiagen, Valencia, CA, USA) with the primers ALK-F and ALK-R (Table S1). The relative copy number ratio of ALK was expressed in comparison with the reference gene LINE1 [39].

Extraction and Characterization of EVs Released from Lung Cancer Cells into Conditioned Medium and Serum
Serum samples from two ALK-translocated lung adenocarcinoma patients were centrifuged at 1500× g at 4 • C for 10 min to remove any cells. The supernatant was passed through 0.22 µm filters to remove contaminating cell debris and large EVs. Conditioned medium (CM) was collected from cells cultured in RPMI-1640 or ACL4 media supplemented with exosome-depleted FBS (System Biosciences, Mountain View, CA, USA) and also filtered through 0.22 µm filters prior to EV isolation. EVs in conditioned media or serum samples were isolated using miRCURY Exosome Isolation Kits (Exiqon, Vedbaek, Denmark) in accordance with the manufacturers' protocols. The isolated EVs were characterized by particle size analysis and immunoblot analysis. The particle size distribution of extracted EVs was examined by NanoSight fluorescent nanoparticle tracking analysis (System Biosciences). The EV protein markers were confirmed by immunoblot analysis with specific anti-CD63 (1:1000, Proteintech, Rosemont, IL, USA), anti-GP96 (1:1000, Invitrogen), and anti-GAPDH (1:10000, Proteintech) antibodies. All relevant data of our experiments were submitted to the EV-TRACK knowledgebase (EV-TRACK ID: EV180001) [40].

EV-RNA Extraction, TaqMan microRNA Assay and Quantitative PCR (qPCR)
To elucidate the role of EV-miRNA in intratumoural heterogeneity, the expression level of 20 cancer-related miRNAs in different FA34-sensitive and -resistant subclones was assayed. The selection of this panel of cancer-related miRNAs was based on their biological relevance to different types of cancer [41,42]. Total EV-RNA was first extracted by TRIzol LS Reagent (Invitrogen). EV-miRNA quantitation was performed using TaqMan microRNA Assays (Applied Biosystems, Foster City, CA, USA). Then DNase-treated EV-RNA was converted to complementary DNA (cDNA) with specific stem-loop primers. All real-time PCR reactions were carried out in a StepOnePlus real-time PCR system (Applied Biosystems). For intracellular miRNAs, the relative expressions of the target miRNAs were normalized to the expressions of small RNA RNU6B. For EV-miRNAs, the relative expressions of the target miRNAs were normalized to the average threshold cycle (Ct) values of all miRNAs determined in the same sample. For quantitation of EV-lncRNAs MEG3 and XIST, EV-RNA was reversely transcribed into cDNA by the QuantiNova Reverse Transcription Kit (Qiagen). The relative expression levels of the EV-lncRNAs were normalized to the expression levels of GAPDH. The relative expression levels of cellular PTEN and MSH2 were quantified by the same method as mentioned above. The sequences of primers used are listed in Supplementary Table S1.

EV Transfer
The effects of EV transfer from resistant subclones to sensitive subclones were assessed. EVs from the different resistant subclones were isolated and transferred to TKI sensitive subclones. Isolated EVs were labelled with 1,1 -dioctadecyl-3,3,3 ,3 ,-tetramethylindodicarbocyanine, 4-chlorobenzenesulfnate salt (DiD) (5 µg/mL) (Invitrogen). Cells were pre-incubated with 5-(N-ethyl-N-isopropyl)-amiloride (EIPA, 50 µM, Tocris Bioscience, Bristol, UK) for 30 min as indicated. Uptake of EVs in recipient cells was confirmed using fluorescence microscopy. To determine transfer of EV cargos, cellular RNA in recipient cells was extracted and the expression of specific miRNA and lncRNA assayed by qPCR as described above. To determine the functional effects of EV transfer, cell viability and motility were determined by MTT cell viability assay and cell migration assay respectively. Cell migration was performed using the transwell migration assay in EV recipient cells [43]. Briefly, ALK-TKI-sensitive subclones FA34.8 were treated with EVs (10 µg/mL isolated from CM of drug resistant subclones for 24 h. Treated cells were then subjected to transwell migration assay with corresponding EV loaded into the lower chamber. After further incubation for 20 h, the migrated cells were fixed and stained; the stain of the migrated cells was eluted for measurement of absorbance at 550 nm.

Serial Circulating EV-RNA Levels in ALK-Translocated Lung Adenocarcinoma Patients on ALK-TKI Treatment
Serial blood samples were taken from two patients with ALK-translocated lung adenocarcinoma along their treatment course. The first patient ( Figure 4A) was a 56 year-old, male, non-smoker with a left upper lobe lung adenocarcinoma with ALK rearrangement confirmed by FISH on endobronchial biopsy. He had mediastinal lymphadenopathy and malignant left pleural effusion. Baseline blood was taken before commencement of crizotinib (Baseline), when the patient had stable disease on crizotinib (Stable disease), and when the patient had disease progression with enlargement of the left upper lobe mass while on crizotinib (Disease progression). The second patient ( Figure 4B) was a 79 year-old, male, non-smoker with a left lower lobe lung adenocarcinoma and malignant left pleural effusion. EML4-ALK translocation was confirmed by endobronchial biopsy and the patient was initially treated with crizotinib but there was clinical disease progression despite three months of crizotinib. Therapy was changed to ceritinib with blood taken (Baseline). The patient showed stable disease (Stable disease) for the next eleven months before the disease progressed again (Disease progression) while on ceritinib.
All blood samples were processed immediately upon collection and the respective serum portions were kept frozen at −80 • C for later analysis. Serum EV isolation, characterization and EV-RNA extraction were performed as described in previous sections. The study was approved by the Institutional Review Board/Ethics Committee of the Hong Kong University/Hong Kong Hospital Authority Hong Kong West Cluster (HKU/HA HKWC IRB/EC UW 16-104).

Statistical Analysis
EV-RNAs and proteins levels are reported as mean ± standard deviation. Normal distribution of quantitative data was confirmed. Between-group comparisons were performed with either Chi-square tests or Student's t-tests where appropriate; multiple group comparisons were performed with one-way ANOVA followed by Tukey's multiple comparisons post hoc. All statistical analyses were performed using GraphPad Prism software 6.01 (GraphPad Software Inc., San Diego, CA, USA). Statistical significance levels were all two-sided and were taken at p < 0.05.

Conclusions
A panel of drug resistant subclones of ALK-translocated lung adenocarcinoma cell lines, namely FA34 and FA121, was established. The established subclones demonstrated a range of different drug sensitivity towards different ALK-TKIs. EV transfer from ALK-TKI-resistant subclones to sensitivity subclones showed potential mediation of drug resistance. Serum EV-miR-21-5p, miR-486-3p, MEG3 and XIST levels may correlate with disease progression and treatment response in patients with ALK-translocated lung adenocarcinoma prescribed ALK-TKI treatment ( Figure 5). This warrants further confirmation in a larger cohort. This study highlighted the urgent need to develop a deeper understanding of intratumoural heterogeneity via EV transfer. The results also imply that EV-RNAs may serve as a novel circulating biomarker for monitoring treatment response and progress of ALK-translocated lung adenocarcinoma.  Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Figure S1: Effects of EVs transfer from ALK-TKI resistant subclones to sensitive subclones on cell migration. Figure S2: Complete expression profiles of serum EV-associated miRNAs in two ALK-translocated patients. Table S1: List of primers used in this study.  Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6694/11/1/104/s1, Figure S1: Effects of EVs transfer from ALK-TKI resistant subclones to sensitive subclones on cell migration. Figure S2: Complete expression profiles of serum EV-associated miRNAs in two ALK-translocated patients. Table  S1: List of primers used in this study.