Circular RNA circPDSS1 promotes osteosarcoma progression by sponging miR-502-3p and miR-4436a.

Osteosarcoma (OS) is a highly aggressive bone cancer. Patients with OS frequently develop drug resistance in clinical treatment, and the prognosis has not been improved significantly. There is an urgent need to identify novel markers and therapeutic targets. In this study, we focused on the highly expressed noncoding circular RNA circPDSS1 in OS, and studied its functional roles and downstream targets in OS cells by CCK-8, clone formation assay, transwell assays. Additionally, we performed luciferase reporter assay, RNA pull-down experiment and qRT-PCR to validate the micoRNA targets of circPDSS1. The involvement of circPDSS1 in tumorigenesis was also investigated in mouse xenografts model. The expression of circPDSS1 was significantly upregulated in OS tissues and cell lines. Patients with high circPDSS1 expression were associated with poorer progression-free survival (PFS) and overall survival (OS) as compared to those with low circPDSS1 expression. CircPDSS1 knockdown significantly inhibited the viability, clone formation ability and invasion ability of OS cells, and induced cell apoptosis, which were associated with the upregulation of proapoptotic proteins and the impairment of prosurvival signaling. Molecular mechanism study further demonstrated that circPDSS1 modulates OS cell functions by regulating the expression of miR-502-3p and miR-4436a. Our data suggest that circPDSS1 acts as a molecular sponge of miR-502-3p and miR-4436a regulates the proliferation and invasion of OS cells and promote the malignant progression of OS.

Introduction

Osteosarcoma (OS) is a malignant tumor in children and adolescents, which is categorized as a highly aggressive bone cancer [1]. OS usually originates from primitive mesenchymal bone cells in long bones, and is characterized of rapid development, metastasis, and resistance to radiotherapy and chemotherapy [2]. At present, the incidence of pediatric OS worldwide is close to 11 million, and this data is still rising [3]. The standard treatment of OS is optimal surgery accompanied by radiochemical therapy [4,5]. However, the prognosis of OS patients has not been improved significantly even with the combined therapy [4]. Therefore, exploring the molecular mechanisms of OS progression and drug resistance holds promise for the identification of therapeutic targets and the formulation of novel therapeutic strategies.

Circular RNA (circRNA) is characterized by covalently closed loops with no protein-coding ability [6,7]. The ring shape makes it highly resistant to enzyme degradation [8]. Although the biological functions of circRNAs have been largely ignored in the past few decades, the wide application of sequencing technology enables researchers to identify numerous circRNAs in various tissues. At the same time, accumulating studies have shown that circRNAs are engaged in diverse functional regulations in cancer, affecting the proliferation, metastasis, differentiation and progression of cancer [9,10]. For example, circRNA ciRS-7 is highly expressed in different types of cancers, and it regulates tumorigenesis through the interaction with miR-7 [11-13]. CircRNA circ-HIPK3 is another well-characterized circular RNA which regulates the proliferation and glycolysis of non-small-cell lung carcinoma (NSCLC) by inhibiting miR-381-3p [14]. In addition, a series of studies identified the dysregulation of multiple circular RNAs in OS. For example, the overexpression of circ_0007534 can accelerate the growth of OS cells by inhibiting apoptosis [15]. Hsa_circ_0002052 upregulation inhibits the proliferation of OS cells by regulating the miR-1205 and attenuating the Wnt/β-catenin pathway [16]. CircLRP6 promotes the development of OS by negatively regulating KLF2 expression [17]. The dysregulation of circular RNAs adds another layer of complexity in the pathogenesis and progression of OS. A previous study showed that circ_100571 (circPDSS1) was highly expressed in the sequencing results of OS samples [18]. However, the functional role of circ_100571 and its molecular mechanisms on OS progression remain unclear.

CircPDSS1 is mapped to chr10: 27024168-27024508 and is derived from the PDSS1 gene locus [18]. The length of circPDSS1 is 195 bp, and its mechanism of action in OS is unknown. In view of this, our study investigated the functional roles and downstream targets of circPDSS1 in OS cells by CCK-8, clone formation assay, transwell assays. We also performed luciferase reporter assay, RNA pull-down experiment and qRT-PCR to validate the micoRNA targets of circPDSS1. The involvement of circPDSS1 in tumorigenesis was also investigated in mouse xenografts model. Our data revealed a regulatory module of circPDSS1-miR-502-3p/miR-4436a in promoting the malignant phenotype of OS cells.

Materials and methods

Sample collection of OS tissues

A total of 60 OS tissue specimens and paracancerous tissue specimens were collected in the Zhuzhou Central hospital. After surgical resection, tissues were stored in a preservation tube and stored in liquid nitrogen. Informed consent of all patients enrolled in the study were obtained. All experiments involving the above-mentioned clinical samples were approved by the ethics committee of Zhuzhou Central Hospital.

Cell culture and transfection

All OS cell lines (U2OS, HOS, Saos2, and MG63) and normal bone cell line hFOB 1.19 used in this study were purchased from the Cell Bank of the Chinese Academy of Sciences. All the above-mentioned cells were cultured with Dulbecco’s modified Eagle medium (DMEM, Hyclone, South Logan, Utah, USA) or F12 Base (Hyclone) containing 10% fetal bovine serum (FBS; Gibco, Grand Island, New York, USA) and 1×penicillin and streptomycin (100 U/mL P/S; Gibco) in a humidified incubator at 37°C and 5% CO2.

Oligonucleotides including miR-502-3p mimic, miR-4436a mimic, microRNA negative control (miR-NC), siRNA for CircPDSS1 (si-circPDSS1), miR-502-3p inhibitor and miR-4436a inhibitor were synthesized by GenePharma (China, Shanghai). si-circPDSS1 sequence: 5′-GTCAGCAGGTTTCATCACACA-3′. In order to overexpress circPDSS1, pcDNA 3.1-circRNA plasmid was constructed. Transfections of the above molecules into cells were performed by Lipofectamine 3000 (Invitrogen, Carlsbad, California, USA) following the manufacturer’s protocol. 4 µg plasmid and 200 nM of miR-345-3p mimic or the NC were used to transfect cells in a six-well plate at 80% confluency. 48 h after transfection, cells were harvested for further experiments.

RNA extraction and qRT-PCR

Total RNA from tissues and cells were extracted using Trizol reagent (15596026, Thermo Fisher Scientific) according to the manufacturer’s instructions. 5 μg of total RNA was used for cDNA synthesis by RevertAid First Strand cDNA Synthesis Kit (K1622, Thermo Fisher Scientific). A 7500 Real-Time PCR System (Applied Biosystems/Life Technologies, Carlsbad, California, USA) was used to perform qPCR analysis using SYBR premix EX TAQ II kit (RR820A, Takara, Dalian, China). The following PCR cycling condition was used: 95°C 2 min, 40 cycles of 95°C 30 s, 60°C 30 s and 72°C 60 s. Finally, qPCR data analysis was performed using the 2−ΔΔCt method and 18S rRNA and U6 gene was used as internal reference for cytoplasmic and nuclear fractions. The primer sequences used in the study are as follows: circPDSS1, forward, 5′-GTGGTGCATGAGATCGCCT-3′; reverse 5′-GGGTTGTGTGATGAAACCTG-3′. 18S, forward, 5′-CCAACCTGGTTGATCC-3′; reverse, 5′-CCTTGTTAACGACTTC-3′. U6, Forward, 5′-AAAGCAAATCATCGGACGACC-3′; reverse, 5′-GTACAACACATTGTTTCCTCGGA-3′. miR-502-3p, forward, 5′-ACACTCCAGCTGGGAAT GCACCTGGGCAAGG-3′; reverse, 5′-CTCAACTGGTGTCGTGGA-3′. miR-4436a, forward, 5′-GCACGACGAGGCGG TGAAGACGGACAGGAC-3′; reverse, 5′-TATGGTTGTTCACGACTCCT-3′.

Nuclear/cytoplasmic fractionation

For nucleoplasm fraction experiment, the nuclear and cytoplasmic faction was extracted using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific, 78833), and the total RNA in each fraction was purified using Trizol reagent (Invitrogen, 15596026) according to the manufacturer’s protocol. An equal number of cells was used for total cell lysate RNA extraction, which serves as the total cellular RNA level control for normalization. The extracted RNA was quantified by RT-qPCR.

Immunoblotting assay

Cells and tissues were lysed using precooled RIPA lysis buffer (Beyotime, Shanghai China), and then centrifuged at 10 000g for 10 min. The concentration of the supernatants containing total proteins was quantified by a BCA Protein assay kit (Beyotime, Shanghai, China). 10 ug protein of each sample was separated by SDS-PAGE electrophoresis and then transferred to the PVDF membrane. The membrane was incubated and blocked with 5% skimmed milk for 1 h, and incubated with the primary antibodies at 4°C overnight. The antibodies and dilution used in this study: Bax (dilution ratio 1:1000), Bcl-2 (dilution ratio 1:1000), cleaved caspase 3 (dilution ratio 1:1000), p-AKT (dilution ratio 1:1000), AKT (dilution ratio 1:1000), p-mTORC1(S2448) (dilution ratio 1:1000), mTORC1 (dilution ratio 1:1000), p-mTORC2(S2481) (dilution ratio 1:1000), mTORC2 (dilution ratio 1:1000), p-P70S6K (dilution ratio 1:1000), P70S6K (dilution ratio 1:1000), p-ERK1/2 (dilution ratio 1:1000), ERK1/2 (dilution ratio 1:1000), p-P38 (dilution ratio 1:1000), P38 (dilution ratio 1:1000), p-JNK1/2 (dilution ratio 1:1000), JNK1/2 (dilution ratio 1:1000), Ki-67 (dilution ratio 1:1000), PCNA (dilution ratio 1:1000) and GAPDH (dilution ratio 1:5000) (All the antibodies were purchased from Abcam, Massachusetts, USA). The membrane was washed 3 times with TBST for 5 min each. After wash, the membrane was further incubated with HRP-linked secondary antibody (1:3000; Cell signaling, Massachusetts, USA) at room temperature for 1 h. Then the membrane was washed four times with TBST and the protein bands were visualized using an enhanced chemiluminescence kit (Santa Cruz, Texas, USA, sc-2048) and photographed on a on a GelDoc XR+ system (Bio-Rad, California, USA). GAPDH was used as the loading control.

Dual luciferase reporter assay

The wild-type (WT) sequences containing the predicted binding sites of miR-502-3p and miR-4436a and circPDSS1, or the mutant (Mut) sequences were cloned into the pmirGLO luciferase reporter (Promega, E1330). Reporter plasmid and Renilla luciferase (hRlucneo) control plasmid were co-transfected into cells with either microRNA mimics or miR-NC using Lipofectamine 3000 reagent. 48 h after transfection, the relative luciferase activities were measured using Dual-Luciferase Reporter Assay Kit (Promega, E1910) on a luminescence microplate reader (Infinite 200 PRO; Tecan). The relative firefly luciferase activity in the reporter plasmid was normalized to that of Renilla luciferase (hRlucneo) control plasmid.

RNA pull-down assay

Cell lysates were collected by IP lysis buffer (Beyotime, Shanghai, China) and were incubated biotinylated circPDSS1 oligo and Control oligos (NC). 10% of the lysates was saved as the input. The mixture was further incubated with M-280 streptavidin magnetic beads (Sigma-Aldrich, 11205D) at 4°C shaking overnight. A magnetic bar was used to pull down the magnetic beads and associated nucleic acids, then the samples were washed four times with high salt wash buffer. Both the input and the elutes from the pull-down were purified with Trizol reagent (Invitrogen) according to the manufacturer’s protocol. Quantitative RT-PCR analysis was performed using to determine the miRNAs precipitated by NC and circPDSS1 oligo.

Cell proliferation assay

Cell proliferation assays were performed using CCK8 kit (Dojindo; Munich, Germany). Cells were seeded into a 96-well plate at a density of 2000 cell/well and cultured in a humidified cell culture incubator for 0, 24, 48, 72 and 96 h, respectively. Subsequently, 10 μL CCK8 reaction solution was added to the cell culture at the indicated time point and incubated for 1 hour in a humidified cell culture incubator. The light absorption value (OD value) in each condition was captured at 450 nm wavelength on a microplate reader.

Apoptosis assay

The detection of cell apoptosis was performed using FITC Annexin V Apoptosis Detection Kit (Invitrogen, San Jose, California, USA) according to the manufacturer’s instructions. In brief, cells with different treatments were trypsinized and resuspended in the staining solution, which contained 5 μL Annexin V-FITC and 5 μL PI in 1000 μL cell resuspension with 1 million cells. Cells were incubated for 30 min in the dark. Stained cells were centrifuged and washed twice with 1 × PBS and resuspended in 400 μL PBS. The percentage of apoptotic cells was detected by BD FACS CantoTM II Flow Cytometer (BD Biosciences).

Colony formation assays

Cells with indicated treatment were trypsinized and resuspended in a culture medium. Cells were seeded into a six-well plate (2000 cells/well) and cultured for 14 days and the culture medium was changed every 3 days during the period. After 14 days, cells were fixed with 4% paraformaldehyde at room temperature for 10 min and stained 0.5% crystal violet (Beyotime, Shanghai, China) for 20 min. Subsequently, the number of colonies was counted and the morphology of the colonies was photographed under Leica AM6000 microscope.

Transwell invasion assay

Transfected cells were starved for 24 h in serum-free medium. Cells were trypsinized and resuspended in serum-free medium. The transwell upper chamber with a pore size of 8 μm (Corning, New York, USA) was coated with Matrigel (Yeason, China). Cells were inoculated into the transwell upper chamber in serum-free medium and 500 μL of 10% serum-containing medium was added to the lower chamber. After 18 h, culture medium was discarded and the cells were fixed with 4% paraformaldehyde at room temperature for 10 mins and stained with 0.5% crystal violet (Beyotime, Shanghai, China) for 20 min. Cells were photographed under Leica AM6000 microscope and the number of invading cells was counted.

Animal model

Four-week-old nude mice were purchased from the Model Biology Center (Shanghai, China). 0.2 mL of cell suspension containing 1 × 107 HOS cells transfected with si-NC or si-circPDSS1 was injected subcutaneously into the lower back of mice (n = 5 per group). After the injection, the tumor volume was measured every week by calipers and recorded until the 5th week (tumor volume = 0.5 × width2 × length). The animals were then sacrificed, and the xenografts were collected and weighed. The xenografts were stored in liquid nitrogen until further experiments. All the animal experimental procedures complied with the guidelines of the Chinese ‘Experimental Animal Management Regulations’. All animal-related protocols were approved by the ethics committee of experimental animals in Zhuzhou Central Hospital.

Data analysis

The data analysis involved in the study was carried out using GraphPad Prism 6.0 software. The statistical difference between the two groups was compared using unpaired student’s t tests. Comparisons among multiple groups were analyzed using one-way analysis of variance (ANOVA) with Tukey’s post hoc test for pairwise comparison. The relationship between the expression of circPDSS1 and the clinical–pathological data of OS was analyzed by chi-square test. Kaplan–Meier Curve and log-rank test were used to compare the survival rates of different groups of patients. Data were reported as the mean ± SD. P < 0.05 was considered to be statistically different.

Results

circPDSS1 is highly expressed in OS tissues and cell lines

In a recent study, Yang et al. performed RNA sequencing on osteosarcoma tissues, and their sequencing results showed that circular RNA circ_100571 (circPDSS1) is differentially expressed in OS tissues and normal tissues [19]. In view of this, we collected 60 pairs of OS and paracancerous tissues to further confirm the expression of circPDSS1 in OS. Our qRT-PCR analysis showed that the expression of circPDSS1 was significantly upregulated in OS tissues as compared with para-cancerous normal tissues (Fig. 1a). Then, we took the median expression value of circPDSS1 as the cut-off to divide 60 OS patients into circPDSS1 low expression group (n = 30) and circPDSS1 high expression group (n = 30). We found that the expression of circPDSS1 was closely related to TNM staging and distant nodal metastasis, but showed no correlation with the age, gender, tumor differentiation, and tumor size of the patients (Table 1). Kaplan–Meier survival curve revealed that the overall survival and progression-free survival (PFS) of the circPDSS1 high expression group were poorer when compared to the low expression group (Fig 1b,c). We further analyzed the expression of circPDSS1 in four OS cell lines (U2OS, HOS, Saos2 and MG63) and human osteoblast line hFOB 1.19. The results indicated that the expression of circPDSS1 was significantly increased in OS cell lines (Fig. 1d).

Fig. 1

CircPDSS1 is highly expressed in OS tissues and cell lines. (a) CircPDSS1 expression was detected by qRT-PCR in 60 pairs of OS tissues and para-cancerous tissues. (b) and (c) Kaplan–Meier analysis of overall survival and progression-free survival in OS patients with high and low circPDSS1 expression. (d) The expression of circPDSS1 in OS cell lines (U2OS, HOS, Saos2 and MG63) and human osteoblast line hFOB 1.19 was analyzed by qRT-PCR (n = 3 experiments). *P < 0.05, **P < 0.01, *** P < 0.001.

Table 1

Correlation of circPDSS1 expression with clinicopathologic features of osteosarcoma

Factor		circPDSS1 expression		P value
		Low (n = 30)	High (n = 30)
Sex				0.216
	Men	19	17
	Women	11	13
Age				0.439
	≤20	20	22
	>20	10	8
Tumor size				0.625
	≤5 cm	18	19
	>5 cm	12	11
Pathological differentiation
Well/moderate	14		12
Poor	16		18
TNM				0.004
	I/II	21	8
	III/IV	9	22
Lymph node metastasis				0.029
	Absence	20	9
	Presence	10	21
Distant metastasis				0.018
	Absence	19	7
	Presence	11	23

circPDSS1 silencing inhibits cell growth and invasion but induces apoptosis in OS cells

In order to further study the role of circPDSS1 in OS cells, we synthesized circPDSS1 specific siRNA (si-circPDSS1) and the scramble control (si-NC) to examine the effect of circPDSS1 knockdown. qRT-PCR analysis showed that compared with si-NC group, si-circPDSS1 significantly reduced the expression of circPDSS1 in HOS and Saos2 cells (Fig. 2a). We further evaluated the effects of circPDSS1 knockdown on cell proliferation, clonogenic ability, invasion ability and apoptosis. Our results showed that knockdown of circPDSS1 significantly suppressed cell proliferation (Fig. 2b), clonal formation ability (Fig. 2c), invasion ability (Fig. 2d), but significantly increased the percentage of apoptotic cells (Fig. 2e). We also analyzed the expression of a variety of proteins involved in apoptosis, cell growth and survival by western blot. The results showed that knockdown of circPDSS1 significantly reduced the phosphorylation level of cell proliferation and survival-related proteins including Akt (T308), ERK1/2, JNK, p38, mTORC1 (S2448), mTORC2(S2481) and p70S6K(T389), while the levels of proapoptotic protein Bax and cleaved caspase-3 were increased (Fig 2f). Together, these data showed that circPDSS1 silencing inhibits cell growth and invasion but induces apoptosis in OS cells.

Fig. 2

CircPDSS1 promotes growth and invasion but inhibits the apoptosis of OS cells. (a) Knockdown efficacy of si-circPDSS1 was validated in HOS and Saos2 by qRT-PCR. (b–e) The effects of circPDSS1 on cells proliferation (b), clonal formation ability (c), invasion ability (d) and cell apoptosis (f) were examined in HOS and Saos2. (f) The protein levels of proteins involved in cell growth, survival and apoptosis were detected by western blot. GAPDH was used as loading control (n = 3 experiments). **P < 0.01, *** P < 0.001.

circPDSS1 serves as a sponge for miR-502-3p and miR-4436a

Different localization of circRNA in the cell determines the different mechanism of actions [6,7]. In order to further clarify how circPDSS1 regulates cell functions, we first investigated the subcellular localization of circPDSS1 by extracting the RNA from nuclear and cytoplasmic fractions. qRT-PCR results showed that circPDSS1 was mainly localized in the cytoplasm (Fig. 3a). On the basis of this, we speculated that circPDSS1 may act as a molecular sponge of miRNAs. We then searched circBank and starbase databases to predict miRNA molecules which potentially interact with circPDSS1. We identified two microRNAs (hsa-miR-4436a and hsa-miR-502-3p) were shared by the prediction results of circBank and starbase databases (Fig. 3b). We next applied miR-4436a and miR-502-3p mimics, which can effectively mimic the overexpression of miR-4436a and miR-502-3p in cells (Fig. 3c,d). To confirm the functional interactions between circPDSS1 and miR-502-3p or miR-4436a, we performed luciferase reporter assay using the wild type (WT) reporter containing predicted binding sequence as well as the mutated reporter (MUT). The results showed that overexpression of miR-502-3p mimic or miR-4436a mimic could inhibit the luciferase activity of WT reporter, and the mutation of the binding site of miR-502-3p or miR-4436a (MUT reporter) abrogated the inhibitory effects (Fig. 3e,f). We further confirmed that circPDSS1 oligonucleotide could pull down both miR-502-3p and miR-4436a by RNA pull-down assay (Fig. 3g). Knockdown of circPDSS1 increased the expression of miR-502-3p and miR-4436a in HOS cells (Fig. 3h). Consistent with this, the expression of miR-502-3p and miR-4436a in OS cells were significantly lower as compared with hFOB 1.19 cells, (Fig. 3I,j).

Fig. 3

CircPDSS1 serves as a sponge for miR-502-3p and miR-4436a. (a) The percentage of circPDSS1, 18s rRNA and U6 transcript in the cytoplasm and nuclear fractions were detected in HOS cells by qRT-PCR. 18S rRNA and U6 genes were used as cytoplasmic and nuclear control. (b) miRNAs targets of circPDSS1 were predicted by circBank and starbase databases. (c, d) The overexpression efficacy of miR-502-3p and miR-4436a mimics were validated in HOS cells by qRT-PCR. (e, f) The functional interaction between circPDSS1 and miR-502-3p or miR-4436a were evaluated by dual-luciferase reporter assay in HOS cells. (g) RNA-pull down assay using biotinylated circPDSS1 oligo and Control oligos (NC). The precipitated RNA was quantified by qRT-PCR. (h) The effects of circPDSS1 knockdown on miR-502-3p and miR-4436a expression in HOS cells were analyzed by qRT-PCR. (i) and (j) Relative expression of miR-502-3p and miR-4436a was detected in OS cells and human osteoblast line hFOB 1.19 by qRT-PCR (n = 3 experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

circPDSS1-miR-502-3p/miR-4436a axis regulates malignant phenotype of OS cells

In order to further confirm the involvement of miR-502-3p and miR-4436a in the function of circPDSS1 in OS cells, we applied miR-502-3p inhibitor and miR-4436a inhibitor to investigate their effect on circPDSS1 silencing, which could significantly reduce miR-502-3p and miR-4436a expression (Fig. 4a,b). Functional experiment showed that miR-502-3p inhibitor or miR-4436a inhibitor partially rescued the inhibitory effect of si-circPDSS1 on cell proliferation (Fig. 4c), clonal formation ability (Fig. 4d) and cell invasion ability (Fig. 4e) and also suppressed apoptosis caused by si-circPDSS1 (Fig. 4f). These results indicate that circPDSS1 regulates cell functions by modulating the expression of miR-502-3p and miR-4436a in OS cells.

Fig. 4

CircPDSS1-miR-502-3p/miR-4436a axis regulates malignant progression of OS cells. (a, b) The effect of miR-502-3p inhibitor and miR-4436a inhibitor on gene expression was examined in HOS cells by qRT-PCR. (c–f) The effects of circPDSS1-miR-502-3p/miR-4436a axis on cell proliferation (c), clonal formation ability (d), invasion ability (e) and cell apoptosis (f) were investigated in HOS cells (n = 3 experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

circPDSS1-miR-502-3p/miR-4436a axis promotes tumorigenesis of OS cells in vivo

In order to study whether the circPDSS1-miR-502-3p/miR-4436a axis plays a similar functional role in vivo, HOS cells were transfected with si-NC or si-circPDSS1 and injected into nude mice. Xenograft tumor size and weight were measured and the results demonstrated that si-circPDSS1 significantly inhibited tumorigenesis in nude mice (Fig. 5a,b). We also performed qRT-PCR and the results showed that the level of circPDSS1 in si-circPDSS1 group xenografts was significantly reduced, while the expression of miR-502-3p and miR-4436a were significantly increased (Fig. 5c). In addition, the protein level of Ki-67 and PCNA (cell proliferation markers) in si-circPDSS1 xenograft was also significantly reduced, indicating a reduced cell proliferation (Fig. 5d). We also examined the expression of proteins involved in apoptosis, cell growth and survival by western blot. Consistent with the in vitro data, the phosphorylation level of cell proliferation and survival-related proteins including Akt (T308), ERK1/2, JNK, p38, mTORC1 (S2448), mTORC2(S2481) and p70S6K(T389) were suppressed in si-circPDSS1 xenograft, while the levels of proapoptotic protein Bax and cleaved caspase-3 were increased (Fig. 5e). Therefore, circPDSS1-miR-502-3p/miR-4436a axis promotes tumorigenesis of OS cells in vivo.

Fig. 5

CircPDSS1-miR-502-3p/miR-4436a axis promotes proliferation of OS cells in vivo. (a) HOS cells transfected with si-NC or si-circPDSS1 were injected subcutaneously into the lower back of mice (n = 5 in each group). (a) Tumor size was measured 5 weeks postinjection. (b) Tumor weight was analyzed 5 weeks postinjection. (c) Relative expression of circPDSS1, miR-502-3p and miR-4436a were detected in si-NC and si-circPDSS1 xenografts by qRT-PCR. (d) The protein levels of Ki-67 and PCNA were detected in si-NC and si-circPDSS1 xenografts by western blot. (e) The expression of signaling pathway proteins involved in cell growth, survival and apoptosis were detected by western blot. GAPDH was used as loading control (n = 3 experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

Discussion

OS is a highly aggressive cancer [19]. Although significant progress has been made in the diagnosis and treatment of OS recently, the prognosis of OS patients has not been significantly improved. A large number of studies have shown that mutations in cancer-related genes contribute to the malignant progression of OS [20]. The wide application of high-throughput sequencing technologies starts to unveil the dysregulation and functional application of various circRNAs in cancer. CircRNAs have the potential to regulate the expression of target genes by acting as miRNA sponges, and they are proposed to serve as a promising biomarker [21].

Although the functions of most circRNAs are not clear, many studies have shown that the dysregulation of certain circRNAs plays an important role in the occurrence and development of OS [15-17]. In this study, we found that circPDSS1 was upregulated in OS tissues, which is consistent with the results of a previous study [18]. In addition, another previous study has revealed that circPDSS1 regulates NEK2 (NIMA related kinase 2) by serving as a sponge of miR-186-5p and promotes the progression of gastric cancer [22]. circPDSS1 is also upregulated in bladder cancer and may play a role in the disease progression by acting as an upstream inhibitor of miR-16 [23]. Although the specific location of circPDSS1 in the above study is different from that of circPDSS1 in our study, they are all derived from the locus of the PDSS1 gene, which implies that different isoforms of circRNAs could derive from the same gene locus. We also found that expression of circPDSS1 was significantly upregulated in OS tissues and cells, and the overall survival of OS patients with high circPDSS1 expression were poorer. circPDSS1 knockdown significantly inhibits cell proliferation and clone formation ability of OS cells and promotes cell apoptosis. We also analyzed the expression of a variety of proteins involved in apoptosis, cell growth and survival by western blot and these data showed that circPDSS1 silencing suppresses cell growth and survival pathways but induces proapoptotic signals.

Our study revealed that circPDSS1 regulates the behavior of OS cells by down-regulating the expression of tumor suppressor miR-502-3p and miR-4436a. The role of miR-502-3p and miR-4436a in OS has not been reported previously. In other cancers, it has been shown that miR-502-3p can inhibit cell proliferation and invasion of hepatocellular carcinoma by targeting SET (SET Nuclear Proto-Oncogene) [24]. Similarly, in gallbladder cancer, miR-502-3p can also promote the progression of gallbladder cancer by negatively regulating SET, which is downstream of LncRNA-HGBC [25]. In addition, circ-RPL15 can promote the progression of gastric cancer by regulating the miR-502-3p/OLFM4/STAT3 signal axis [26]. Our results showed that miR-502-3p is downregulated by circPDSS1 in OS, and miR-502-3p inhibitor suppressed OS cell proliferation and invasion. Similarly, miR-4436a is another downstream molecule regulated by circPDSS1, and its low expression promotes the proliferation and invasion of OS cells. For miR-4436a, a previous study has shown that miR-4436a can form RNA-induced silencing complexes and promote SNHG11-mediated CTNNB1 expression in a ceRNA mode, thereby hindering the development of lung cancer [27]. Although this study reported the interaction of miR-4436a with lncRNA, it has a similar regulatory mechanism as circRNA. Our data suggest that circPDSS1 functions as the sponge of miR-4436a and it also regulates the expression of miR-4436a in tumor tissues. Interestingly, our study indicates that circPDSS1 may function as an upstream inhibitor for both miR-502-3p and miR-4436a in OS cells. The luciferase reporter assay and RNA pull-down assay further support that circPDSS1 physically interacts with miR-502-3p and miR-4436a. However, whether the interaction between circPDSS1 and miR-502-3p or miR-4436a is OS-specific needs to be further clarified.

In conclusion, this study revealed that circPDSS1 may regulate the behavior of OS cells by downregulating the dual signals of tumor suppressors miR-502-3p and miR-4436a, thereby promoting the malignant progression of OS.

Acknowledgements

This work was supported by the Scientific Research Project of Hunan Health and Family Planning Commission C2017074.

S.T. and Z.Y. designed the experiments; S.T., X.T. and Z.J. performed the experiments; Q.H. and L.W. helped with clinical studies, data collection; S.T. wrote the manuscript; Z.Y. did the manuscript editing; All authors reviewed the manuscript and approved the submitted version.

Conflicts of interest

There are no conflicts of interest.