Introduction

Osteosarcoma (OS) is the most common malignant tumor of bone and mainly affects children and adolescents [1]. The annual incidence of OS is 4–5 cases per 1,000,000 individuals worldwide [2]. Despite development of OS treatment approaches, such as chemotherapy and curative resection, OS overall survival is poor [3]. The molecular mechanism underlying OS development remains unclear [4]. Therefore, in-depth exploration of the mechanism underlying OS development is important to develop effective preventions and therapies.

Circular RNAs (circRNAs) are recently identified molecules classified into a type of noncoding RNAs. CircRNAs show diverse biological functions, including cellular proliferation, invasion and metastasis [5]. CircRNAs function by serving as miRNA sponges and subsequently regulating target gene expression [6]. Several studies have explored the role of circRNAs in OS. Fang et al. [7] demonstrated that circ_0000337 is significantly upregulated in OS cell lines and promotes OS progression via the miR-4458/BACH1 pathway. Another study conducted by Zhang et al. [8] revealed that has_circ_0002052 stimulates the progression of OS through sponging miR-382. To date, however, the potential mechanisms of circRNAs in OS progression have rarely been studied.

MicroRNAs (miRNAs) are noncoding RNAs that mainly function as inhibitors of target genes [9]. MiRNAs affect the biological behaviors of multiple cancers, including OS. Liu et al. [10] revealed that miR-874-3p inhibits OS cell line migration through targeting RGS4. Further studies on miRNA and OS progression are required. In recent years, a number of tumor-related databases have been established, including Gene Expression Omnibus (GEO) and TCGA.

In this study, differential gene expression analyses involving OS samples were performed using GEO datasets. We validated one circRNA, named Circ-XPR1, generated from exons 8 and 9 of the XPR1 gene. Furthermore, we demonstrated that Circ-XPR1 facilitates OS cell proliferation by sponging miR-214-5p to regulate DDX5 expression.

Methods

Analysis of GEO expression dataset

The following three datasets from the GEO expression dataset were obtained from the GEO database (https://www.ncbi.nlm.nih.gov/geo/, Access date: 15 April 2020): GSE96964, GSE65071, and GSE12865. The GSE96964 dataset contains circRNA expression profiles, including one human osteoblast hFOB1.19 cell line and seven human osteosarcoma cell lines (U2OS, U2OS/MTX300, HOS, MG63, 143B, ZOS, and ZOSM). The GSE96964 platform consisted of the Agilent-069978 Arraystar Human CircRNA Microarray V1. The GSE65071 dataset contains miRNA expression profiles, containing 20 OS samples and 15 healthy controls. The GSE65071 platform consisted of the Exiqon Human V3 MicroRNA PCR Panel I + II. The GSE12865 dataset contains 12 OS samples and 2 normal human osteoblasts (HOB) samples, and the GSE12865 platform consisted of the Affymetrix Human Gene 1.0 ST Array platform.

Background correction and quantile normalization were performed for the GSE96964, GSE65071, and GSE12865 datasets using the Affy package ( https://www.bioconductor.org/) [11] and robust multiarray average (RMA) [12].

Identification of differentially expressed circRNAs, miRNAs, and mRNAs among these datasets was performed using the Limma algorithm in R [13]. For the differentially expressed analysis, the thresholds were p value < 0.05 and logFC > 1 (upregulated) or logFC <  − 1 (downregulated).

To identify the targeting miRNAs of Circ-XPR1, we intersected differentially expressed miRNAs from the GSE65071 dataset, Targetscan database (https://www.targetscan.org), and miRanda database (https://www.microrna.org). Using the same method, we analyzed the miR-214-5p target gene using the intersection of GSE12865, Targetscan, miRbase, and RNAhybrid to obtain a final list of genes.

Human tissues

In total, 20 primary OS tissues were acquired from OS patients at Henan Provincial People’s Hospital between May 2011 and May 2013. All samples were confirmed by pathological diagnosis after surgery. Normal bone specimens were obtained from tissues adjacent to osteosarcoma tissues. Informed consent was obtained from all patients, and the study procedure was approved by the Ethics Committee of Henan Provincial People’s Hospital. For survival analyses, the overall survival of patients was divided into low and high expression groups according to the median expression of the gene of interest, and it was presented using a Kaplan–Meier plot.

Cell culture

Two OS cell lines (MG-63 and U2OS) were obtained from ATCC Company (Manassas, VA, USA) and maintained in a humidified incubator at 37˚C/5% CO2. MG-63 and U2OS cells were cultured in Dulbecco's Modified Eagle Medium (DMEM, HyClone, Thermo-Fisher Scientific, Ottawa, ON) containing 10% fetal bovine serum, 100 U/ml penicillin G, and 100 U/ml streptomycin.

RNA isolation of nuclear and cytoplasmic fractions

Nuclear and cytoplasmic fractions of MG-63 cells were extracted using the Nuclear and Cytoplasmic RNA Purification Kit (Cat. 21,000, Norgen, USA) according to the manufacturer’s instructions. RNA extraction from nuclear and cytoplasmic fractions was performed using TRIzol reagent (Invitrogen).

Assessment of half-life of cells

Transcription was stopped by treatment with actinomycin D (10 μg/ml), and RNA was extracted at various time points (0, 4, 8, 12 and 24 h) following actinomycin D treatment. Relative expression level of target genes was detected by qRT-PCR.

RNA fluorescent in situ hybridization (FISH)

The Circ-XPR1 (5′-TCCCAGTGACAGTCGAATAG-3′) probe labeled with FAM was designed and synthesized by RiboBio Company (Guangzhou, China). Nuclei were stained with DAPI. The FISH experiment was performed using the BersinBio circRNA FISH Kit. Images were acquired with a confocal fluorescence microscope (Olympus Imaging, Tokyo, Japan).

Plasmid construction

The Circ-XPR1 overexpression plasmid was designed and purchased from GenePharma (Shanghai, China) and inserted into a pcDNA3.0 vector. Empty vector (pcDNA) was used as the control. Circ-XPR1 siRNA was purchased from GenePharma (Shanghai, China). The sequences of Circ-XPR1 siRNA1 and 2 were as follows: siRNA1, 5′-AACTGCTGCTGTATTTAAACT-3′; and siRNA2, 5′-GCTGCTTAAACTGCTGCTGTA-3′. The sequences of miR-214-5p mimic and negative control miRNA (miR-NC) were as follows: miR-214-5p mimic, 5′-ACTCCACUAGCAAACTUGAA-3′; and negative control sequence, 5′-TCAUGACGAAUGCTAGGCC-3′. siRNAs were transfected into cells using Lipofectamine 2000 Transfection Reagent (Life Technologies).

RNase R assay and Real-time PCR

Harvested tissues or cells were placed in TRIzol® Reagent (Invitrogen). RNA samples were divided into two portions as follows: one portion was used for circRNA expression detection; and the other portion was used for GAPDH and miRNA expression measurement. Reverse transcription was performed using SuperScript III Reverse Transcriptase (Invitrogen). To purify circRNAs, 2 μg of RNA was digested with 3 U/mg Rnase R (Geneseed, Guangzhou, China) for 20 min at 37 °C. Real-time PCR analysis was performed with a PrimeScriptTM RT-PCR Kit (TaKaRa Bio Inc., Japan) on an ABI Prism 7000 sequence detection system in duplicate on two different plates. U6 or GAPDH was used as the endogenous control. The following primer sequences were used: Circ-XPR1: forward 5′-TGAAGAAATCGGTGCTGTGG-3′, reverse 5′- TGGTATCTCGGTTGATGTAGGT-3′; XPR1: forward 5′-GGCTGCGTCCTCCGACTC-3′, reverse 5′-GGGGAGAGATCACGATGTTGT-3′; miR-214-5p: forward 5′-ACACTCCAGCTGGGGACAAT-3′, reverse 5′-CTCAACTGGTGTCGTGGAGTCG-3′; miR-665: forward 5′-GGAAGAGAACTTTGCCGTTGAA-3′, reverse 5′-GTGACGAGCCATTTCCTCCTT 3′. U6: forward 5′-CTCGCTTCGGCAGCACA-3′; reverse 5′-AACGCTTCACGAATTTGCGT-3′; and GAPDH: forward 5′-ACAACTTTGGTATCGTGGAAGG-3′; reverse 5′-GCCATCACGCCACAGTTTC-3’.

Western blot assay

Total protein was extracted with RIPA lysis buffer (Beyotime, Shanghai, China) containing phenylmethanesulfonyl fluoride (PMSF) to measure DDX5 protein expression. Protein concentration was evaluated by a BCA Protein Assay Kit (Beyotime) following the manufacturer’s protocol. Each protein sample (25 µg) was separated on a 12% SDS-PAGE gels, transferred onto a PVDF membrane (BioRad, USA), incubated with 5% no-fat milk, incubated with primary antibodies (DDX5, 1:1000, Abcam; and GAPDH, Abcam, 1:1000) overnight at 4˚C, and blocked with secondary antibodies at room temperature for 2 h. Finally, ECL detection system reagents (Millipore, MA, USA) were used to visualize the protein blots. ImageJ (National Institutes of Health, Bethesda, MD) was used to analyze the gray intensity of the bands.

Colony formation assay

For the colony formation assay, transfected cells (1000 cells/well) were seeded into 6-well plates. After 1 week of cultivation, cells were fixed with 75% alcohol for 1 h and then stained with 0.1% crystal violet staining solution (Solarbio, Beijing, China). Colonies were quantified using ImageJ software (https://rsb.info.nih.gov/ij/).

Cell Counting Kit-8 assay

Transfected cells were divided into the following three groups: NC group (cells transfected with scramble siRNA), siRNA-1 group (cells transfected with siRNA-1 targeting Circ-XPR1) and siRNA-2 group (cells transfected with siRNA-2 targeting Circ-XPR1). The CCK-8 assay (Solarbio Biotechnology, Beijing, China) was performed to measure cell viability at different time points (1, 2, 3, 4 and 5 days). In brief, an equal number of cells per group was seeded into 96-well plates. The CCK-8 solution (10 µL per well) was then added and incubated at 37 °C for 1 h, and absorbance was measured at 450 nm using a spectrophotometer (Olympus, Tokyo, Japan). The CCK-8 assay was performed in triplicate, and each experiment was repeated three times.

Luciferase reporter assay

MG-63 or U2OS cells were transfected with a mixture containing 50 ng of firefly luciferase reporter, 5 ng of pRL-CMV Renilla luciferase reporter, and 5 pmol miR-214-5p mimic in a 96-well plate and incubated for 48 h. Luciferase activity was measured using the Dual-luciferase Reporter Assay System (Promega, Madison, Wisconsin, USA). Luciferase activity was normalized against the activity of Renilla luciferase.

Statistical analysis

All analyses were conducted using SPSS 15.0 statistical software (IBM, SPSS, USA). Data were analyzed by unpaired Students t tests for two groups and ANOVA for more than two groups. Correlation analysis between miR-214-5p and Circ-XPR1/DDX5 was performed using Pearson’s correlation analysis. Overall survival (OS) and disease-free survival (DFS) were calculated using Kaplan–Meier survival analysis. p values less than 0.05 were considered statistically significant.

Results

Identification of Circ-XPR1 by GEO datasets

We first identified differentially expressed circular RNAs according to an online dataset (GSE96964) containing one human osteoblast hFOB1.19 cell line and seven human osteosarcoma cell lines. Based on fold change ≥ 1 and p < 0.05, eight circular RNAs were upregulated, whereas 103 circular RNAs were downregulated (Fig. 1a, b). We validated this data by quantitative reverse transcription-PCR (qRT-PCR) and found that the hsa_circ_0005909 (circ-XPR1) expression level was significantly increased by approximately 3.0-fold in OS tissue compared to control tissue (p < 0.05, Fig. 1c).

Fig. 1
figure 1

Identification of the Circ-XPR1 circular RNA using the GSE96964 dataset. a Heatmap of differentially expressed circRNAs between one human osteoblast hFOB1.19 and seven human osteosarcoma cell lines. Green and red represent downregulation and upregulation, respectively. b Volcano plots for differentially expressed circRNAs. The red, green and black dots represent downregulation, upregulation and no significance, respectively. c Relative expression of Circ-XPR1 in control and OS tissues. d Abundance of Circ-XPR1 and linear XPR1 mRNA treated with RNase R or MOCK as assessed by RT-PCR. e Half-life of Circ-XPR1 and XPR1 after treatment with actinomycin D for 0, 4, 8, 12 and 24 h in MG-63 cells. f FISH assay and sublocalization analyses were performed to determine the cellular location of Circ-XPR1. g Kaplan–Meier curves of overall survival (OS) and disease-free survival (DFS) of 20 OS patients with low or high Circ-XPR1 expression. h Schematic diagram demonstrating that exons 8 and 9 of XPR1 constitute Circ-XPR1. *p < 0.05, Values represent mean ± SD, n = 3 independent experiments

Full size image

Circ-XPR1 showed a strong resistance to digestion by RNase R, whereas the linear RNA of XPR1 was highly degraded (Fig. 1d), indicating that circ-XPR1 is a circRNA. We then used actinomycin D to inhibit transcription of cellular RNA. The abundance of circ-XPR1 was relatively stable until 24 h after treatment with actinomycin D, while XPR1 was decreased by twofold after 4 h and nearly disappeared within 24 h (Fig. 1e). RNA FISH indicated that circ-XPR1 localized in the cytoplasm, suggesting that circ-XPR1 may regulate gene expression by functioning as a miRNA sponge (Fig. 1f). Total RNA was isolated from nuclear (N) and cytoplasmic (C) fractions, and both circ-XPR1 and XPR1 were preferentially localized within the cytoplasm. Taken together, these results confirmed that circ-XPR1 a stable cytoplasmic circRNA derived from the XPR1 gene.

Overall survival and disease-free survival were affected by abnormal expression of circ-XPR1 in OS. In OS patients, circ-XPR1 was upregulated and negatively correlated with overall survival and disease-free survival (p = 0.004 and p = 0.025, respectively; Fig. 1g). According to the genomic structure, exons 8 and 9 in the XPR1 gene are flanked and form the circ-XPR1 on chromosome 1q25.3 (Fig. 1h). Unlike circ-XPR1 expression, the XPR1 gene was not significantly changed in GSE12865 (Supplement S1 A), which was further verified by RT-PCR (Supplement S1 B).

Circ-XPR1 promotes OS cell proliferation

To further explore the biological function of circ-XPR1 in OS cell proliferation, we designed two circ-XPR1 oligonucleotides to target the back-splice junction sites (Fig. 2a). MG-63 and U2OS cells transfected with circ-XPR1 siRNA-1 and siRNA-2 showed a significant decrease in cell viability compared to those transfected with control siRNA as assessed by the Cell Counting Kit-8 (CCK-8) assay 4 and 5 days after transfection (Fig. 2b).

Fig. 2
figure 2

Circ-XPR1 promotes the proliferation of OS cells. a Schematic representation of the sequence around the back-splice junction site of Circ-XPR1 and the siRNAs targeting the junction site (siRNA-1 and siRNA-2). b CCK-8 assays were performed to examine the cell proliferation rate of MG63 and U2OS cells after knockdown of Circ-XPR1 by siRNA-1 and siRNA-2 at 1, 2, 3, 4 and 5 days. c Relative abundance of Circ-XPR1 and XPR1 expression in MG-63 cells after transfection with Circ-XPR1 siRNA-1 and siRNA-2. d Colony formation assays showed decreased colony formation ability after Circ-XPR1 knockdown in MG-63 and U2OS cells. e Relative abundance of Circ-XPR1 in MG-63 cells after transfection with Circ-XPR1 plasmids. f Cell proliferation was measured by CCK-8 at 1, 2, 3, 4 and 5 days after transfection of Circ-XPR1 and control plasmids. g Colony formation assays showed decreased colony formation ability after overexpression of Circ-XPR1 in MG-63 and U2OS cells. *p < 0.05, Values represent mean ± SD, n = 3 independent experiments

Full size image

Circ-XPR1 siRNA-1 and siRNA-2 inhibited the relative abundance of circ-XPR1 by onefold, while the expression of the XPR1 linear species was not affected (Fig. 2c). Colony formation assay was used to assess cell proliferation ability, and proliferation was indicated by the number of colonies. The colony formation capacity was reduced in MG-63 and U2OS cells after silencing circ-XPR1 expression (siRNA-1 and siRNA-2; p < 0.05; Fig. 2d).

For circ-XPR1 expression analysis, a circ-XPR1 overexpression vector was constructed and transfected into MG-63 cells via Lipofectamine 2000. Circ-XPR1 was confirmed by abundance analysis (Fig. 2e). CCK-8 analysis demonstrated that MG-63 and U2OS cells overexpressing circ-XPR1 showed increased viability (Fig. 2f). Colony formation assays were consistent with the CCK-8 assays (Fig. 2g). Taken together, these data suggested that circ-XPR1 is a novel circRNA that promotes the proliferation of OS cell lines.

Circ-XPR1 acts as a sponge of miR-214-5p

To identify a novel potential target for Circ-XPR1, we first analyzed the GSE65071 dataset, which includes 20 OS samples and 15 healthy controls. The total number of differentially expressed miRNAs was 164 (78 upregulated miRNAs and 86 downregulated miRNAs; Fig. 3a, b).

Fig. 3
figure 3

Circ-XPR1 acts as a miR-214-5p sponge in OS cells. a GSE65071 was used to evaluate differentially expressed miRNAs between OS and control samples. The significant genes are shown by a volcano plot with a threshold of fold change ≥ 1 and p value < 0.05. The red, green and black dots represent the downregulation, upregulation, and no significance, respectively. b Heatmap of the differentially expressed miRNAs between OS and control samples. Green and red represent downregulation and upregulation, respectively. c Two miRNAs (miR-214-5p and miR-665) were identified by Venn analysis of GSE65071, Targetscan and miRanda databases. d Relative abundance of miR-214-5p and miR-665 in MG-63 cells after transfection with siRNA-1 or siRNA-2 and NC. e Schematic illustration of the Circ-XPR1- and miR-214-5p-binding sites. f The potential miR-214-5p-binding site of Circ-XPR1-WT. g Relative luciferase activity was analyzed for WT or Mut 3′-UTR XPR1 reporter plasmids cotransfected with miR-214-5p mimic or miR-NC. h Relative expression of miR-214-5p after transfection of Circ-XPR1 plasmids and control plasmid. i Relative expression of miR-214-5p in OS and control tissues. j Correlation between Circ-XPR1 and miR-214-5p in OS tissues (R2 = 0.435, p = 0.002). *p < 0.05, Values represent mean ± SD, n = 3 independent experiments

Full size image

By overlapping miRNAs targeting Circ-XPR1 in the Targetscan and miRanda databases with miRNAs differentially expressed in the GSE65071 dataset, two overlapping miRNAs, miR-214-5p and miR-665, were identified (Fig. 3c). After knockdown Circ-XPR1 expression by two siRNAs, miR-214-5p and miR-665 expression levels were measured. miR-214-5p expression was markedly increased after transfection of siRNA-1 and siRNA-2 (7.85-fold), but miR-665 expression was not significantly increased (1.22-fold increase, Fig. 3d). Circ-XPR1/miR-214-5p interactions were then predicted using circMir (version 1.0), and the targeting site between Circ-XPR1 and miR-214-5p is shown in Fig. 3e.

In total, we identified 17 putative miR-214-5p-binding sites in Circ-XPR1 (Fig. 3f). MiR-214-5p significantly decreased luciferase activity of WT XPR1 3′UTR but not Mut XPR1 3′UTR (mutant) in MG-63 and U2OS cells (Fig. 3g). Overexpression of circ-XPR1 in MG-63 cells decreased miR-214-5p expression compared to the control group (p < 0.05, Fig. 3h).

The relative expression of miR-214-5p was decreased in OS tissues (n = 20) compared to control tissues (n = 20, Fig. 3i). Spearman’s correlation analyses revealed significant negative correlations between circ-XPR1 and miR-214-5p in OS tissues (R2 = 0.435, p = 0.002, Fig. 3j). These findings indicated that circ-XPR1 may act as a sponge to adsorb miR-214-5p.

miR-214-5p Targets DDX5 in OS Cells

To identify target genes regulated by miR-214-5p, we performed a differentially expressed gene analysis using the GSE12865 dataset, which contains 14 samples (12 osteosarcoma tumor samples and 2 normal human osteoblasts as controls). The array data was background adjusted and normalized using the mean-centering approach in R software (Fig. 4a). We generated a volcano plot (Fig. 4b) and heat map (Fig. 4c) showing differential gene expression between control and OS samples. We overlapped three miRNA target gene databases (Targetscan, https://www.targetscan.org/mamm_31/; miRBase, https://www.mirbase.org/ and RNAhybrid, https://bibiserv.techfak.uni-bielefeld.de/rnahybrid/) and one dataset from the GEO database (GSE12865). Four genes, namely, DDX5, CPA4, NRG1, and CCBE1, were identified as shown in Fig. 4d.

Fig. 4
figure 4

DDX5 is a target of miR-214-5p in OS cells. a Comparison of the expression value before normalization and after normalization of GSE12865. b Volcano plot of differentially expressed genes in GSE12865. The red, green and black dots represent downregulation, upregulation, and no significance, respectively. c Heat map displaying differential mRNA expression in OS within the GSE12865 dataset. Color scale at right side of heatmap shows the expression level. Red indicates upregulation, and green indicates downregulation. d Venn diagram demonstrates the intersection genes from GEO dataset (GSE12865), Targetscan, miRbase, and RNAhybrid. e Prediction of the binding sites between miR-214-5p and DDX5. f Relative luciferase activity following cotransfection of miR-214-5p mimic with DDX5 3′UTR WT or MUT in MG-63 and U2OS cells. g Relative expression of DDX5 in miR-214-5p mimic and miR-NC groups in MG-63 and U2OS cells. h Relative expression of DDX5 in MG-63 and U2OS cells after transfection with Circ-XPR1 siRNA-1 and siRNA-2. i Relative expression of DDX5 in OS and control tissues. j Correlation analysis between Circ-XPR1 and DDX5 (R2 = 0.348, p = 0.006). *p < 0.05, values represent mean ± SD, n = 3 independent experiments

Full size image

The free energy value of miR-214-5p to its target binding site ranged from -15.34 to − 38.9 kcal/mol, and the binding free energy for DDX5 was maximum. Thus, DDX5 was selected for further investigation. Putative target genes of miR-214-5p were predicted by bioinformatics analysis using miRNA target prediction software (Targetscan, https://www.targetscan.org/mamm_31/). Furthermore, a dual-luciferase reporter assay revealed that DDX5 is a putative target of miR-214-5p (Fig. 4e).

The dual-luciferase reporter assay revealed that miR-214-5p decreased the luciferase activity of WT-DDX5 but did not affect the luciferase activity of MUT-DDX5 in MG-63 and U2OS cells (Fig. 4f). Furthermore, we measured the expression of DDX5 in the miR-214-5p and negative control groups. As shown in Fig. 4g, the expression of DDX5 rapidly decreased in response to miR-214-5p, but miR-214-5p NC expression did not decrease significantly in MG-63 and U2OS cells. After transfection of two siRNAs targeting circXPR1, DDX5 expression was significantly decreased in MG-63 and U2OS cells (Fig. 4h). The expression level of DDX5 was significantly upregulated in OS tissues compared to control tissues (p < 0.05, Fig. 4i). Furthermore, circXPR1 and DDX5 were also positively correlated (R2 = 0.348, p = 0.006, Fig. 4j). Western blot assay revealed that miR-214-5p decreased DDX5 protein expression in MG-63 and U2OS cells (Fig. 5a). Knockdown of circXPR1 also decreased DDX5 protein expression in MG-63 and U2OS cells (Fig. 5b). Furthermore, DDX5 protein expression was upregulated in OS tissues compared to control tissues (Fig. 5c). These findings indicated that the DDX5 gene is the target of miR-214-5p.

Fig. 5
figure 5

DDX5 protein expression in OS tissues and cells. a Western blot assay assessed DDX5 protein expression in miR-214-5p-transfected OS cells. b DDX5 protein expression in OS cells transfected with circ-XPR1 siRNA. c DDX5 protein expression in OS and control tissues. *p < 0.05, values represent mean ± SD, n = 3 independent experiments

Full size image

Discussion

In this study, we first identified circ-XPR1 as a key upregulated circRNA involved in OS cell proliferation. In addition, gain- and loss-of-function experiments confirmed that circ-XPR1 promotes OS cell proliferation. For the first time, we identified and verified the role and potential function of circ-XPR1 in OS progression. Circ-XPR1 is derived from exons 8 and 9 within the XPR1 gene locus. XPR1 is a 696-amino acid protein and has been identified to play a vital role in tumor progression [14]. In this study, we used a GEO dataset to investigate differentially expressed circRNAs in OS and identified a novel circRNA, namely, Circ-XPR1. Circ-XPR1 was upregulated in OS tissue and associated with shorter overall survival rates. Subcellular fractionation assay and RNA FISH showed that Circ-XPR1 mainly accumulated in the cytoplasm, suggesting that Circ-XPR1 might act as a sponge of miRNAs to regulate gene expression and OS progression.

We used a GEO dataset to identify, differentially expressed miRNAs, and two putative target miRNAs of Circ-XPR1 were selected by the intersection of two databases and the GEO dataset. Circ-XPR1 contains a conserved miR-214-5p target site that was validated by a dual-luciferase reporter gene assay, miRNA abundance analysis and correlation analyses. miR-214-5p has multiple biological functions, including inhibition of proliferation in nonsmall lung cancer [15], cervical cancer [16] and esophageal cancer [17]. The results of this study were similar to those of previous studies demonstrating that miR-214-5p is downregulated in OS tissues and may possess antitumor effects [18, 19]. Zhang et al. [18] demonstrated that miR-214-5p targets ROCK1 and suppresses proliferation and invasion of human OS cells. Moreover, Zhou et al. [19] performed in vivo and in vitro studies, which demonstrated that LINC00612 functions as a ceRNA for miR-214-5p to promote OS cell invasion.

To identify the targets of miR-214-5p, we preprocessed the gene expression profiles collected from the GEO database (GSE12865), resulting in 2839 genes. Four putative target mRNAs of miR-214-5p were selected by the intersection of three databases and GSE12865. A luciferase reporter assay revealed that miR-214-5p inhibited DDX5-WT activity, suggesting that DDX5 is directly targeted by miR-214-5p.

DDX5, a nuclear prototypic member of the DEAD box family, is abnormally expressed in a variety of malignancies, including neuroblastoma [20], breast cancer [21], prostate cancer [22] and colorectal cancer [23]. In addition, silencing of DDX5 by siRNA blocks the migration of carcinoma cells [24]. Chen et al. [25] reported that lncRNA DLEU1 sponges miR-671-5p and upregulates DDX5 expression to aggravate OS carcinogenesis, suggesting that DDX5 promotes the occurrence and development of OS.

The expression level of DDX5 in OS tissue samples was significantly increased and positively correlated with Circ-XPR1 expression. Therefore, we propose that the Circ-XPR1/miR-214-5p/DDX5 axis is a novel target for OS progression. However, these results do not preclude the possibility that other potential target genes are involved in OS progression. Future investigations should aim to further reveal the regulatory role of the CircXPR1/miR-214-5p/DDX5 axis in OS in vivo.

Conclusion

In summary, we identified a novel circRNA derived from the XPR1 gene. We demonstrated that Circ-XPR1 acts as tumor promotion factor through regulating the miR-214-5p/DDX5 axis in OS. Our study demonstrated that Circ-XPR1 may be a potential target for therapeutic intervention of OS.