LncRNA FENDRR sensitizes doxorubicin-resistance of osteosarcoma cells through down-regulating ABCB1 and ABCC1

Long noncoding RNAs (LncRNAs) act as crucial regulators in various cancers including osteosarcoma (OS), yet their potential roles and molecular mechanisms in OS chemoresistance remain unclear. In the present study, we investigated the role and potential regulatory mechanism of the most down-regulated expressed lncRNA, FENDRR screened by our previous lncRNA microarray analysis between the paired doxorubicin-resistant and sensitive human osteosarcoma cell lines (MG63/DXR vs MG63). FENDRR expression was down-regulated in the doxorubicin-resistant OS cell lines and tissues and negatively correlated to the poor prognosis of OS patients. Overexpression of FENDRR suppressed doxorubicin-resistance, G2/M phase of cell cycle, and promoted cell apoptosis of osteosarcoma cells in vitro and tumor growth in vivo whereas FENDRR knockdown had the opposite effects. In addition, we found that FENDRR was mainly located in the cytoplasm and could regulate the drug resistance of osteosarcoma cells by negatively affecting posttranscriptional expression of ABCB1 and ABCC1. Together, our study demonstrated that lncRNA FENDRR may act as an inhibitory molecule of doxorubicin-resistance through down-regulating the expression of ABCB1 and ABCC1 genes in osteosarcoma cells. These findings may extend the function of FENDRR in tumor progression and provide a novel target for reversing OS chemoresistance.


INTRODUCTION
Osteosarcoma (OS) is the most common type of solid bone tumors in children and adolescence, which is aggressive, high-grade, with lung metastases in about 20% patients at first diagnosis [1]. Although the systemic therapy including neoadjuvant chemotherapy and surgical excision has raised the 5-year survival rate of OS patients to about 70% within the past decades, the occurrence of intrinsic or acquired drug-resistance has greatly hindered the survival rate with further improvement [2,3]. Moreover, there is no established second-line chemotherapy for OS patients who are resistant to the commonly used chemotherapeutics including doxorubicin, cisplatin, methotrexate, and ifosfamide [4,5]. Thus, it is crucial to understand the molecular mechanisms underlying OS chemoresistance to develop more effective treatments against this disease [6].
Long noncoding RNA (lncRNA) is a type of non-protein coding RNA transcripts longer than 200 nucleotides [7]. Rapidly accumulating evidence has demonstrated that lncRNAs could play important roles in a variety of cellular physiological and pathologic processes, including cell proliferation, apoptosis, autophagy, invasion, metastasis, chemoresistance, radioresistance and so on [8,9]. Actually, several lncRNAs including H19 [10,11], MALAT1 [12,13], UCA1 [14,15], HOTTIP [16], have been reported to be involved in the chemoresistance of various kinds of malignant tumors, such as bladder cancer, pancreatic cancer, glioblastoma, and so on. However, the expression and specific role of lncRNA in OS chemoresistance have not been well understood.
In the present study, we identified the function and regulatory mechanism of lnRNA FENDRR in OS doxorubicin-resistance, which was screened in our

Research Paper
Oncotarget 71882 www.impactjournals.com/oncotarget previous study on the lncRNA and mRNA expression profiles of the doxorubicin-resistant human OS cell line MG63/DXR and its parental cell line MG63 ascertained by microarray analysis [17]. We found that lncRNA FENDRR was the most down-regulated lncRNA in the doxorubicin-resistant osteosarcoma cells and possibly acted as a drug-resistance suppressing molecule through inhibiting ABCB1 and ABCC1 expression. Besides, our data demonstrated that FENDRR could act as a biomarker to predict the chemotherapy sensitivity and prognosis of OS patients. Thus, these results elucidated a novel role for FENDRR in reducing OS chemoresistance and suggested a potential target to sensitize doxorubicin-resistance in OS.

LncRNA FENDRR was down-regulated in the doxorubicin-resistant osteosarcoma cell lines and tissues verified by microarray and qPCR
In our previous study, LncRNA microarray was conducted in three pairs of doxorubicin-resistant human osteosarcoma cell line MG63/DXR and its parental cell line MG63 to identify the candidate lncRNAs involved in the OS chemoresistance [17]. Among the 3,465 differently expressed lncRNAs, 1,704 lncRNAs were down-regulated more than 2-fold change in the MG63/ DXR cells relative to MG63 and lncRNA FENDRR was the most down-regulated of 22-fold change, which was selected as a candidate for further examination via RT-qPCR ( Figure 1A). To further validate the results of microarray, we analyzed FENDRR expression in three paired doxorubicin-resistant and sensitive human osteosarcoma cell lines (MG63/DXR vs MG63, KH-OS/ DXR vs KH-OS, U2-OS/DXR vs U2-OS) and other two doxorubicin -sensitive osteosarcoma cell lines (SaoS2 and HOS) by qRT-PCR and the results showed that FENDRR expression was significantly desregulated with a dramatically suppressed expression in the doxorubicinresistant cell lines compared with the doxorubicinsensitive cell lines ( Figure 1B). We also evaluated the expression level of FENDRR in 80 osteosarcoma tissues that were divided in the chemoresistant group (N = 40) and chemosensitive group (N = 40). FENDRR was significantly down-regulated in tumor tissues resected from the multidrug resistant osteosarcoma patients and upregulated in the multidrug sensitive patients in accordance with the above findings ( Figure 1C, P < 0.001).We further examined whether FENDRR expression correlated with osteosarcoma prognosis. Kaplan-Meier survival analysis showed that patients with low FENDRR expression had a lower rate of overall life compared with the high FENDRR expression ( Figure 1D, p < 0.01, log-rank test). These results suggested that lncRNA FENDRR was possibly related to the occurrence of chemoresistance and to predict the chemoresponse and prognosis of OS patients.

FENDRR increased doxorubicin sensitivity of osteosarcoma cells in vitro
To further identify the role of lncRNA FENDRR in OS chemoresistance, CCK-8 assay and colony formation analysis were performed to examine the effect of FENDRR on cell sensitivity and proliferation to anticancer drugs of doxorubicin in the MG63 (or KH-OS) and MG63/DXR (or KH-OS/DXR) cells. MG63/DXR (or KH-OS/DXR) cells were transfected with the FENDRR expression vector or empty vector. MG63 (or KH-OS) cells were stably transfected FENDRR-specific siRNA and control siRNA served as a negative control (NC). Satisfactory transfection efficiency was obtained at 48 h post-transfection and confirmed by qPCR (Figure 2A and 2B). IC50 values for doxorubicin in response to FENDRR up-or down-regulation were measured. The results showed that compared with MG63/DXR (or KH-OS/DXR) cells transfected with empty vector, the IC50 values of doxorubicin in MG63/DXR (or KH-OS/DXR) cells transfected with FENDRR were reduced by 50.5% (or 37.5%) (P < 0.01, P < 0.01, Figure 2C). Compared with MG63 (or KH-OS) cells transfected with si-NC, the IC50 values of doxorubicin in MG63 (or KH-OS) transfected with si-FENDRR were increased by 150% (or 120%) (P < 0.01, P < 0.01, Figure 2D).
Similarly, colony formation assays revealed that cell proliferation was significantly suppressed in FENDRR-overexpressing MG63/DXR (or KH-OS/ DXR) cells compared with empty vector transfected cells exposed to 16 μg/mL doxorubicin for 14 days (P < 0.01, P < 0.01, Figure 2E). Cell proliferation was significantly increased in FENDRR-down-regulated MG63 (or KH-OS) cells compared with NC-transfected cells exposed to 4 μg/mL doxorubicin for two weeks (P < 0.01, P < 0.01, Figure 2F). Taken together, these data suggested that FENDRR might resensitize OS cells to doxorubicin in vitro.

FENDRR suppressed cell cycle and promoted apoptosis of osteosarcoma cells
Furthermore, we used flow cytometry analyses to assess the effect of FENDRR on cell cycle and apoptosis. The rates of G0/G1 and S phases were more than in the forced expression of FENDRR of MG63/DXR (or KH-OS/DXR) cells compared with the vector transfected cells whereas FENDRR knockdown in the MG63 (or KH-OS) cells significantly increased the rate of G2/M phase relative to the si-NC group ( Figure 3A and 3B). In addition, compared with negative controls, overexpression of FENDRR resulted in increased apoptosis proportion of MG63/DXR (or KH-OS/DXR) cells when exposed to doxorubicin (0 μg/mL or 8 μg/mL,) for 48 h ( Figure 3C). Likewise, down-regulation of FENDRR caused a decrease in apoptosis proportion of MG63 (or KH-OS) cells www.impactjournals.com/oncotarget compared with the si-NC group treated with doxorubicin (0 μg/mL or 4 μg/mL) for 48 h ( Figure 3D).

Overexpression of FENDRR improves the vivo sensitivity of osteosarcoma cells to doxorubicin
To further assess the underlying effect of FENDRR in suppressing the chemoresistance of osteosarcoma cells to doxorubicin in vivo, we inoculated nude mice with MG63/DXR cells stably transfected with FENDRR expression vector or empty vector or MG63 cells transfected with si-FENDRR or si-NC. These cells were subcutaneously injected into mice followed by treatment with doxorubicin after tumor formation (50 mm 3 ). As shown in the figure, tumors derived from pcDNA-FENDRR-transfected MG63/DXR cells grew significantly  (E) Cell proliferation in response to anticancer drugs was examined using colony formation analysis. Cell colony formation in the FENDRR group of MG63/ DXR (KH-OS/DXR) cells was reduced compared with the FENDRR-NC group when exposed to 16 ug/ml doxorubicin for 14 days. (F) Cell colony formation in the si-FENDRR group of MG63 (KH-OS) cells was increased compared with the si-NC group when exposed to 4 ug/ml doxorubicin for two weeks.*P < 0.05; **P < 0.01.
Oncotarget 71885 www.impactjournals.com/oncotarget slower than those from controls following doxorubicin treatment. However, tumors formed from si-FENDRR transfected MG63 cells grew faster compared with those derived from si-NC transfected cells after doxorubicin exposure ( Figure 4A). Mice were sacrificed after 5 weeks for drug injection. Growth curves of tumor volumes and nude mice weight indicated that si-FENDRR-treated group increased significantly and FENDRR transfected group decreased markedly ( Figure 4B and 4C). Together, these data suggested that dysregulated lncRNA FENDRR might be associated with the doxorubicin-resistance of osteosarcoma cells in vivo.

FENDRR was mainly located in the cytoplasm and regulated the expression of the classical drug resistance related genes of ABCB1 and ABCC1
To examine the subcellular localization of FENDRR, Cy3-labeled probes specific for FENDRR were used for RNA-FISH. This analysis confirmed FENDRR-specific staining was observed in the cytoplasm of MG63 and KH-OS cells, whereas nearly no staining was observed in the nucleus, indicating that FENDRR was predominantly localized in the cytoplasm ( Figure 5A). Then RT-qPCR of nuclear and cytoplasmic fractions of the two cells further validated that FENDRR was mainly located in the cytoplasm ( Figure 5B), indicating its role of regulating the gene expression at the posttranscriptional level.
To clarify the underlying mechanism of lncRNA FENDRR regulating chemoresistance in OS, we further conducted bioinformatics analysis and co-expression network analysis among FENDRR and the reported multidrug resistance-related genes. GO and Pathway analysis were performed among the FENDRR related differently expressed genes and the results of GO analysis revealed that these dysregulated transcripts were associated with oxidation-reduction process (ontology: biological process), mitochondrion (ontology: cellular component) and protein binding (ontology: molecular function) ( Figure 5C). Pathway analysis based on the KEGG database showed that 20 pathways were significantly enriched among the dys-regulated mRNAs, including the "PI3K-Akt signaling pathway" and "NF-kappa B signaling pathway" that have been previously reported to be involved in the occurrence of drug resistance in OS ( Figure 5D).
Besides, we found a negative correlation between FENDRR and ABCB1 (or ABCC1, ABCC3, EGFR, IL-6) with the correlation coefficient less than −0.99. Subsequently, the expression of ABCB1, ABCC3, EGFR and IL-6 was evaluated in MG63 (or KH-OS) cells with FENDRR knockdown and MG63/DXR (or KH-OS/DXR) cells with FENDRR overexpression by qRT-PCR and WB. Interestingly, we found that the mRNA and protein expression level of ABCB1 and ABCC1 decreased, when FENDRR was overexpressed. Meanwhile, ABCB1 and ABCC1 expression increased when FENDRR was knockdown in the two cells, whereas the expression of the other three genes was almost unchanged ( Figure 5E and 5F). These data indicated that lncRNA FENDRR might regulate the sensitivity of osteosarcoma cells to chemotherapy drugs by negatively affecting posttranscriptional expression of ABCB1 and ABCC1.

DISCUSSION
Osteosarcoma is a high-grade malignant bone tumor [18]. Although the introduction of chemotherapy has greatly improved the survival rate of OS patients, the clinical effectiveness is limited by the occurrence of multi-drug resistance, which ultimately leads to poor prognosis [19]. Recently, many lncRNAs have been demonstrated to regulate chemoresistance in various kinds of cancers through affecting the cell cycle, cell apoptosis or some related drug-resistant genes or signal transduction pathways [20][21][22][23][24]. For example, Si X et alreported that lncRNA H19 confers chemoresistance in ERα-positive breast cancer through epigenetic silencing of the proapoptotic gene BIK [10]. Chen J et alfound that lncRNA CCAT1 acts as an oncogene and promotes chemoresistance in docetaxel-resistant lung adenocarcinoma cells through sponging let-7c and further up-regulating Bcl-xl [25]. Li Z et al reported that lncRNA HOTTIP promotes progression and gemcitabine resistance by regulating HOXA13 in pancreatic cancer [16]. Shen et al reported that lncRNA PVT1 epigenetically silences miR-195 and modulates EMT and chemoresistance in cervical cancer cells [26]. Özeş AR et al found that NF-κB-HOTAIR axis links DNA damage response, chemoresistance and cellular senescence in ovarian cancer [27].
Actually, there have been few lncRNAs reported to be involved in the osteosarcoma chemoresistance. We first identified the lncRNA expression profiles of the doxorubicin-resistant human osteosarcoma cell line MG63/DXR and its parental cell line MG63 by microarray analysis [17] and found that lncRNA ODRUL was the most up-regulated of 20-fold change in the doxorubicinresistant OS cell lines, promoted doxorubicin resistance in OS by increasing the expression of transcription factor FOXC2 through a RNA-RNA double-strand structure in the overlapping region, further facilitating ABCB1 expression [28,29].Besides, Wang Y et al reported that lncRNA LINC00161 sensitizes osteosarcoma cells to cisplatin-induced apoptosis by regulating the miR-645-IFIT2 axis [30]. Li Z et al reported that overexpression of lncRNA HOTTIP increases chemoresistance of osteosarcoma cell by activating the Wnt/β-catenin pathway [31]. Zhou Q et al found that lncRNA PVT1 promotes osteosarcoma development by acting as a molecular sponge to regulate miR-195 [32].
In the current study, we focused on the function and mechanism of another lncRNA, FENDRR, also screened by our microarray analysis previously described. www.impactjournals.com/oncotarget As is known to us, FENDRR is located at chr3q13.31, divergently transcribed from the FOXF1 promoter and coexpressed with FOXF1 [33]. Grote et al first identified that Fendrr, homologous to FENDRR in human, is essential for proper heart and body wall development in mouse and it could bind to both polycomb repressive complex 2 (PRC2) and Trithorax group/MLL protein complexes (TrxG/MLL), which play pivotal roles in the control of chromatin structure and gene activity [33]. He further found that Fendrr functions in the lateral mesoderm via epigenetic modification of regulatory elements, thereby adjusting the expression level of target genes in this tissue and setting long-term marks, which allow for the proper control of Fendrr target genes in the descendants of lateral mesoderm cells [34]. Besides, Szafranski reported that FENDRR was expressed in the lung, and associated with a lethal lung developmental disorder called ACD/MPV (alveolar capillary dysplasia with misalignment of pulmonary veins) [34,35]. Apart from its involvement in the heart and body wall or mesoderm development, its role in tumor progression also has been identified in the gastric cancer and non-small cell lung cancer. Xu et al reported that lncRNA FENDRR was down-regulated in gastric cancer cell lines and cancerous tissues and low FENDRR expression predicted poor prognosis. Histone deacetylation was involved in the down-regulation of FENDRR in gastric cancer cells and FENDRR overexpression suppressed invasion and migration by down-regulating FN1 and MMP2/MMP9 expression [36]. Besides, lncRNA FENDRR, also named as lncRNA FOXF1-AS1, was an intergenic lncRNA, next to the FOXF1 gene in the genome location. Miao et al found that loss of lncRNA FOXF1-AS1 regulates epithelial-mesenchymal transition, stemness and metastasis of non-small cell lung cancer cells via EZH2 and down-regulation of FOXF1 [37].
However, there has been seldom report about the FENDRR and osteosarcoma or chemoresistance. In the study, we first identified that FENDRR was consistently down-regulated in the doxorubicin-resistant OS cell lines and tissues and negatively correlated with poor prognosis of OS patients. Functional analysis revealed that overexpression of FENDRR suppressed doxorubicin resistance, G2/M phase of cell cycle, and promoted cell apoptosis of osteosarcoma cells in vitro and tumor growth in vivo whereas FENDRR knockdown had the opposite effects. These results suggested that lncRNA FENDRR may functions as a suppressor in the OS chemoresistance.
As is known to us, the most classical and important molecular mechanism of multidrug resistance in cancer is dysregulation of ABC membrane transporters [38][39][40][41][42]. ABCB1, also named MDR1, is a well-known multidrug resistance-associated gene and P-glycoprotein (P-gp), the protein encoded by ABCB1, is an ATP-dependent  Obviously, the expression of ABCB1 and ABCC1 other than EGFR, IL-6 and ABCC3 greatly decreased in the FENDRR overexpression group and significantly increased in the FENDRR knockdown group relative to the NC groups as confirmed by both PCR and WB detection.*P < 0.05.
Oncotarget 71889 www.impactjournals.com/oncotarget drug efflux pump (especially for xenobiotic compounds) with broad substrate specificity, leading to resistance to anticancer drugs [43]. Besides, ABCC1, also named MRP1, involved in P-gp up-regulation and contributed to mediate drug resistance [44,45]. In fact, several reports have identified that lncRNA could regulate the chemoresistance by influencing the expression of ABCB1 or ABCC1. Wang et al found that lncRNA MRUL, located 400kb downstream of ABCB1, was significantly up-regulated in two multidrug-resistant GC cell lines and MRUL might positively affect the expression of ABCB1 in an orientation-and position-independent manner [28,46]. Xiao et al showed that lncRNA UCA1 functions as a competitive endogenous (ceRNA) of ABCB1 through completely binding the common miR-16 in chronic myeloid leukemia cells [47]. Moreover, Zhang et al reported that lncRNA PVT1 could promote the development of MDR through regulation of mTOR/HIF-1a/P-gp and ABCC1 signaling pathway in gastric cancer cells [48].
In our present study, we found that the expression of FENDRR and ABCB1 or ABCC1 was significantly negatively correlated through the bioinformatics analysis. Further validation by PCR and WB found that FENDRR could negatively regulate the mRNA and protein of ABCB1 and ABCC1 instead of the other genes, such as ABCC3, EGFR or IL-6. On basis of the previously reported mechanism of FENDRR in regulating gene expression, it may promote the expression of ABCB1 and ABCC1 through binding PRC2 to epigenetically modify the gene promoters. However, the specific regulatory mechanism between lncRNA FENDRR and the two drugresistance related genes needed to be further studied by RNA Pull down and chromatin immunoprecipitation.
In summary, our study showed that FENDRR is dramatically down-regulated in the doxorubicin-resistant OS cell lines and tissues and negatively correlated with poor prognosis of OS patients. Moreover, overexpression of FENDRR has the effect of suppressing doxorubicin resistance, G2/M phase of cell cycle, and promoted cell apoptosis of osteosarcoma cells in vitro and tumor growth in vivo. Then lncRNA FENDRR possibly acts as an inhibitory molecule in the OS chemoresistance by down-regulating the posttranscriptional expression of the classical multidrug resistance-related ABCB1 and ABCC1 genes. Our results demonstrated an important lncRNA involved in OS chemoresistance, thereby providing a promising option for facilitating the development of chemotherapy.

Cell lines and culture conditions
MG63, SaoS2 and HOS human osteosarcoma cell lines (American Type Culture Collection (ATCC)) were cultured in DMEM (Gibco, Cat. No. 12491-015, CA, USA) supplemented with 10% fetal bovine serum (Gibco, Gran Island, NY, USA). The doxorubicin-resistant osteosarcoma cell line MG63/DXR, which was kindly provided by Dr. Yoshio Oda [49] (Kyushu University, Fukuoka, Japan), was generated in a stepwise manner by exposing drug-sensitive MG63 cells to increasing doses of doxorubicin (DXR). The paired KH-OS and KH-OS/DXR were kindly donated by Dr. Gonos ES [50] (National Hellenic research Foundation, Athens, Greece) using the same selection method. The paired U2-OS and U2-OS/DXR were kindly donated by Dr. Duan ZF [51] (Massachusetts General Hospital, Boston, USA) and generated by the same method. The surviving cells were subsequently maintained in the conditioned medium with 1 μg/mL DXR (Sigma) to retain its drug-resistant phenotype.

Clinical samples and histological response evaluation
A total 80 of patient's specimens used in this study were described in our previous study. Written informed consent was obtained from all patients at the Shanghai Tenth Hospital between 2006 and 2015.The clinical parameters of osteosarcoma patients in this study are presented in Table 1.

Drug-resistance assay
In vitro drug cytotoxicity was measured by Cell Counting Kit-8 (CCK-8) assays. The cells incubated without drugs were set at 100% survival and were used to calculate the concentration of each cytostatic drug lethal to 50% of the cells (IC50). The ranges of drug concentrations were based on earlier studies and aimed at obtaining an

Colony formation assay
MG63 and KH-OS cells (500 cells per well) were seeded in a six-well plate and cultured for 14 days after treatment. Colonies were then fixed with 10% formaldehyde for 10 min and stained for 5 min with 0.5% crystal violet. The number of visible colonies was counted manually.

RNA extraction and qRT-PCR
Total RNA from all tissues and cells was isolated using Trizol regent. The relative gene expression was calculated by using 2−ΔΔCt method. Primers used for amplifying specific genes in this study are presented in Table 2.

Cell cycle and apoptosis analysis
Cells for cell cycle analysis were stained with propidium iodide using the BD Cycle test plus DNA Reagent Kit (BD Biosciences). Cells for apoptotic analysis were double-stained with Annexin V-FITC and propidium iodide, 48 h after the transfection and were analyzed using a flow cytometer (BD Biosciences) equipped with Cell Quest software (BD Biosciences). The methods used in this assay have been described previously [28].

RNA-fluorescence in situ hybridization (FISH)
Cy3-labeled FENDRR and DAPI-labeled U6 probes were obtained from Genepharma (Shanghai, China). RNA FISH were performed using fluorescent in situ hybridization kit according to the manufacturer's protocol (Thermo Fischer).

Cell cytoplasm/nucleus fraction isolation
Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, USA) were employed to prepare cytoplasmic and nuclear extracts from MG63 and KH-OS cells. RNAs extracted from each of the fractions were subjected to following RT-qPCR analysis to demonstrate the levels of nuclear control transcript (β-actin), cytoplasmic control transcript (U6), and lncRNA FENDRR.  in the four groups (Department of Central Laboratory, Shanghai Tenth Hospital). Tumor growth was examined every other day with a vernier caliper and nude mice weights were also documented simultaneously. Tumor volumes were calculated using the equation: V = A * B 2 /2 (mm 3 ), where A is the largest diameter, and B is the perpendicular diameter. When the average tumor size reached approximately 50 mm 3 , 5.0 mg/kg doxorubicin was administered through tail vein injection every other day. After 5 weeks for drug injection, all mice were killed, necropsies were performed, and tumors were excised. Intratumoral injection of Ad-si-FENDRR (2 × 10 9 plaque-forming units [PFU]) or Ad-si-NC was performed. Mice were photographed before sacrifice with anIVIS@ Lumina II system (Caliper Life Sciences, Hopkinton, MA) beginning 10 minutes after an intratumoral injection of 4.0 mg of luciferin (Gold Biotechnology, Inc., St. Louis, MO) in 100 µL of saline.

Western blot analysis
Protein samples were separated on SDS-PAGE gels and transferred to PVDF membranes (Bio-Rad). PVDF membranes were incubated with primary antibody, then incubated with horseradish peroxidase conjugated secondary antibody. After washing 3 times for 10 min in TBST, the proteins were visualized with an ECL detection system. ImageJ software (BD, Franklin Lakes, NJ, USA) was used to quantify the level of protein expression by calculating integrated density value (IDV) using GAPDH as a control. Antibodies against ABCB1, ABCC1, ABCC3, EGFR, or IL-6 were purchased from Cell Signaling Technology (Danvers, MA, USA).

Bioinformatics analysis
GO analysis was applied to analyze the main functions of the different expression genes related to the FENDRR according to the GO database (www. geneontology.org). Pathway analysis was used to find out the significant pathways of the different genes related to the FENDRR according to KEGG database (Kyoto Encyclopedia of Genes and Genomes, www.genome.jp/ kegg). Co-expressed analysis was conducted between lncRNA FENDRR and its related coding genes with Pearson correlation coefficients equal to or greater than 0.999. The correlation coefficients of FENDRR and the multi-drug resistance genes were shown in the Table 3.

Statistics
All statistical analyses were performed using SPSS 22.0 software (IBM). Data are expressed as the Mean ± SD for at least three independent experiments. Differences between groups were analyzed using the Student's t test or one-way ANOVA when more than two groups were compared. Overall survival was calculated by Kaplan-Meier survival analysis and compared using the log-rank test. A value of P < 0.05 was considered statistically significant.