Tissues of OS and non-cancerous regions were collected from 40 patients (age range, 16–54 years; mean age, 35±1.2 years; male/female ratio, 3:5) by surgery at the 970th hospital of the PLA Joint Logistic Support Force between August 2017 and June 2021. No chemotherapy or radiotherapy was applied to treat the patients before surgery. The tissue samples were verified by two pathologists. All patients were informed of the research design and written informed consent was provided by all patients. The present study was approved (approval no. EC-H-2021-12-16) by the Research Ethics Committee of the 970th hospital of the PLA Joint Logistic Support Force (Yantai, China), which was in accordance with the ethical standards formulated in the Helsinki Declaration (28).

The 293T cell line was purchased from the American Type Culture Collection, and four OS cell lines (HOS, SJSA-1, MG63 and U2OS) and the normal human osteoblast cell line (hFOB1.19) were obtained from ScienCell Research Laboratories, Inc. Cells were cultured in RPMI-1640 medium (HyClone; Cytiva) under a condition of 5% CO₂ and 37°C. All media were supplemented with 10% (v/v) fetal bovine serum (FBS; MilliporeSigma), 100 IU/ml penicillin and 100 mg/ml streptomycin. Overexpression (OE)-vector, small interfering (si)-NC, si-circBLNK, OE-GPX4, inhibitor control, miR-188-3p inhibitor, mimic control, and miR-188-3p mimics were provided by Shanghai GeneChem Co., Ltd. Transfection of cells with a total of 20 nM of each indicated construct was implemented at 37°C for 15 min using Lipofectamine^® 3000 reagent (Invitrogen; Thermo Fisher Scientific, Inc.). After 48 h, the transfected cells were harvested to perform subsequent experiments. The sequences used were as follows: si-NC: 5′-TTCTCCGAACGTGTCACGT-3′; si-circBLNK#1: 5′-AGATCTGTCTGAGAAGAAACA-3′; si-circBLNK#2: 5′-GATCTGTCTGAGAAGAAACAA-3′; miR-NC mimic: 5′-GUCGUAUCCAGUGCAGGGUCC-3′; miR-188-3p mimic: 5′-CUCCCACAUGCAGGGUUUGCA-3′; miR-NC inhibitor: 5′-UUCUCCGAACGUGUCACGUTT-3′; miR-188-3p inhibitor: 5′-GAGGGUGUACGUCCCCAAACGU-3′.

Cytoplasmic and nuclear RNAs were separated using a cytoplasmic and nuclear RNA separation kit according to the manufacturer's protocol. Total RNAs were isolated by TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) from OS tissues and MG63 and U2OS cell lines according to the manufacturer's protocol. For RNase R assay, RNase R (3 U/mg; cat. no. ab286929; Abcam) was applied to the RNAs and incubated for 15 min at 37°C. Afterwards, the remaining RNAs were assessed by reverse transcription-quantitative PCR (RT-qPCR).

Following RNA isolation by TRIzol^® reagent from MG63 and U2OS cells, cDNA was synthesized by a High-Capacity cDNA Reverse Transcription kit (cat. no. 4368814; Thermo Fisher Scientific, Inc.) according to the manufacturer's instruction. The Thermal Cycler Dice System (Takara Bio, Inc.) with the SYBR Green PCR Master Mix (Thermo Fisher Scientific, Inc.) were used to perform qPCR. The reaction mixture was incubated at 95°C for 2 min. The thermocycling program consisted of mainaining at 94°C for 2 min, followed by 30 cycles of 30 sec at 94°C, 30 sec at 56°C, and 60 sec at 72°C. The 2^−ΔΔCq method was employed to assess relative expression levels (29). The primers are listed in Table I.

The protein was extracted from MG63 and U2OS cells as previously described (30). Lysate samples (40 µg) were applied to a 10% SDS-PAGE gel, and following electrophoresis, the separated protein was transferred to a PVDF membrane (Invitrogen; Thermo Fisher Scientific, Inc.). After treatment with 5% skim milk at 37°C for 2 h, the membrane was treated with anti-GPX4 (1:1,000; cat. no. ab125066; Abcam) and GAPDH (1:1,000; cat. no. 92310; Cell Signaling Technology, Inc.) at 4°C for 24 h. Subsequently, the membrane was treated with the IgG H&L (HRP-conjugated; 1:10,000; cat. no. 7076S; Cell Signaling Technology, Inc.) at 37°C for 2 h. The blot was finally developed by ECL reagents (Amersham; Cytiva), and the band intensity was measured using ImageJ version 4.1 (National Institutes of Health).

The nuclear and cytosolic fractions of MG63 and U2OS cells were separated using the Cytoplasmic & Nuclear RNA Purification Kit (100) (cat. no. NGB-37400; Norgen Biotek Corp.) according to the manufacturer's protocol. The RNA was extracted from the nuclear and cytosolic fractions. Afterwards, the subcellular location of circBLNK in OS cells was inferred by its abundance in different fractions.

Cell proliferation was assessed by the CCK-8 assay. In short, following the transfection of MG63 and U2OS cells with corresponding constructs for 72 h, the cells were collected and seeded into a 96-well plate (1×10⁴ cells/well). After 0, 24, 48 and 72 h of cell culture, CCK-8 reagent (Thermo Fisher Scientific, Inc.) (10 µl/well) was applied and incubated for 2 h at 37°C. The 450 nm absorbance was read by a microplate reader (Bio-Rad Laboratories, Inc.).

The MTT assay was used to measure cell growth capability in a 96-well plate. In short, 1×10⁵ MG63 and U2OS cells were seeded in each well. To assess the cell viability, the cells were incubated for 24 h at 37°C. Subsequently, MTT solution (20 µl/well, 5 mg/ml) (Thermo Fisher Scientific, Inc.) was applied and incubated at 37°C for 4 h. Subsequently, dimethyl sulfoxide (DMSO) (150 µl/well) (Invitrogen; Thermo Fisher Scientific, Inc.) was added in order to dissolve the purple formazan. Finally, the absorbance at 570 nm was calculated by an ELISA browser (BioTek EL 800; Omega Bio-Tek, Inc.).

The cells transfected with corresponding constructs for 6 h were seeded into a six-well plate (1×10⁴ cells/well). Then, MG63 and U2OS cells were cultured in complete DMEM supplemented with 10% FBS at 37°C and 5% CO₂. The culture medium was replaced at 2-day intervals. After 14 days of incubation, the colonies were washed twice with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde for 20 min. Next, 4% triformol (Invitrogen; Thermo Fisher Scientific, Inc.) was applied to fix the cells followed by the staining with 0.1% crystal violet (Beijing Solarbio Science & Technology Co.) for 20 min. Finally, the colony number with >50 cells was counted under a stereomicroscope (Olympus Corporation).

Cell apoptosis of MG63 and U2OS cells was assessed by the annexin V-FITC/PI Apoptosis Detection Reagent (cat. no. 88-8102-72; Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. A Beckman Coulter FACSCalibur (BD Biosciences) flow cytometer coupled with the FlowJo 7.6.1 software system (FlowJo LLC) was employed to detect apoptotic cells.

The circRNA expression data was downloaded from Gene Expression Omnibus (GEO; accession no. GSE140256; URL: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE140256). The binding sites between circBLNK and miR-183-3p, and the target genes of miR-183-3p were predicted using CircInteractome (URL: http://circinteractome.irp.nia.nih.gov/circular_rna.html), MicroRNA Target Prediction Database (miRDB; http://mirdb.org/) and TargetScan 7.1 database (www.targetscan.org).

CircBLNK wild-type (circBLNK-WT), circBLNK mutant (circBLNK-Mut), GPX4 wild-type (GPX4-WT), and GPX4 mutant (GPX4-Mut) were subcloned into the pGL3-basic vector (Shanghai GeneChem Co., Ltd.). Subsequently, these constructs were co-transfected into respective 293T cell lines with mimics of miR-183-3p or control using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). After 72 h, the transfected cells were harvested, and the luciferase activity of the cells was calculated by the Luciferase Reporter Assay System (Invitrogen; Thermo Fisher Scientific, Inc.). Luciferase activity was first normalized to that of Renilla luciferase from the same sample.

The RNA pull-down assay was used to validate the relationship between circBLNK and miR-188-3p. Briefly, the MG63 and U2OS cells were transfected with circBLNK probe or NC probe for 48 h at 4°C. A total of ~1×10⁷ cells were dissolved in the soft lysis buffer plus 80 U/ml RNasin (Promega Corporation). The probes were incubated at 4°C for 2 h with 50 µl C-1 magnetic beads (Thermo Fisher Scientific, Inc.) to generate probe-coated beads and then incubated at 4°C for 4 h with 0.7 ml RIPA lysis buffer (Thermo Fisher Scientific, Inc.) to lyse MG63 and U2OS cells. After washing in the lysis buffer (150 mM KCl, 25 mM Tris pH 7.4, 5 mM EDTA and 0.5% NP40), RNA complexes were subjected to centrifugation at 11,100 × g for 10 min and then eluted by denaturation in 1X protein loading buffer for 10 min at 100°C. Finally, miR-142-3p or miR-142-5p transcripts were measured by RT-qPCR.

The RIP assay was used to assess the binding of miR-188-3p to GPX4. After lysis in RIP the MG63 and U2OS cells extract was incubated with magnetic beads conjugated with anti-Ago2 (1:2,000; cat. no. ab186733; Abcam) or normal anti-IgG (1:2,000; cat. no. ab186733; Abcam) at 4°C for 24 h. Next, the magnetic beads were treated with proteinase K, and the enrichment of miR-188-3p or GPX4 in immunoprecipitated RNAs was determined with RT-qPCR.

Ferroptosis analysis was used to measure the ferroptosis cell death. In short, a mixture of 5 µM erastin (Sigma-Aldrich; Merck KGaA), 1 µM ferrostatin-1 (cat. no. 347174-05-4; Sigma-Aldrich; Merck KGaA), 25 µM Z-VAD-FMK (cat. no. V116; Sigma-Aldrich; Merck KGaA), and 20 µM necrosulfonamide (cat. no. ab143839; Abcam) was used to treat MG63 and U2OS cells at 37°C for 24 h. Afterwards, 0.1% DMSO (Invitrogen; Thermo Fisher Scientific, Inc.) was applied to the cells followed by a 4 h incubation at 37°C under serum-free conditions. Finally, the cell viability was examined by the MTT assay.

Levels of total iron and Fe²⁺ in MG63 and U2OS cells were assessed by Iron Assay kit (cat. no. ab83366; Abcam). Levels of MDA and lipid ROS were determined using Lipid Peroxidation Assay Kit (cat. no. ab118970; Abcam) and Cellular ROS Assay Kit (cat. no. ab186027; Abcam), respectively. The aforementioned kits were used according to each manufacturer's protocol.

MitoSOX™ Red (Thermo Fisher Scientific, Inc.) was applied to measure the production of mitochondrial superoxide in MG63 and U2OS cells. First, the working solution was prepared by diluting 5 mM MitoSOX reagent stock solution (cat. no. M36005; Thermo Fisher Scientific, Inc.) with Hanks' balanced salt solution (HBSS; Invitrogen; Thermo Fisher Scientific, Inc.). Subsequently, the working solution was applied to cells and incubated in darkness at 37°C for 10 min. The fluorescence intensity at FL2 was measured by FACSCalibur (BD Biosciences).

A Mitochondrial Membrane Potential Kit (cat. no. ab113850; Abcam) was used to measure the mitochondrial membrane potential in MG63 and U2OS cells according to the manufacturer's protocol.

Subcutaneous injection was performed by Vital River Laboratories. MG63 cells (5×10⁶) in 100 µl PBS were injected into the breast fat pads of four-week-old (weight, 15–20 g) female BALB/c nude mice (n=12). The mice were raised under a 12/12-h light/dark cycle at 18–22°C and 50–60% humidity. Drinking water was continuously provided and food was supplemented three times a week. Both the research team and the veterinary staff monitored animal health and behavior twice daily. Tumor size was measured every three days from one week up to five weeks post-injection. Tumor volume was calculated based on the formula V=0.5² × ab², where a represents tumor length and b indicates tumor width. The humane endpoint about this experiment was set to that the maximum size the tumors allowed to grow in the mice before euthanasia was 2,000 mm³. At the end of animal experiment, the maximum tumor weight/mice weight observed was 7.4%. The maximum tumor diameter observed in this study was 16.7 mm. The duration of the experiment was five weeks. In strict accordance with the principles of animal welfare, anesthesia was used to relieve pain during euthanasia under the euthanasia by mask inhalation of isoflurane vaporized and sacrificed by cervical dislocation. Isoflurane was used at 2% for induction of anesthesia and at 1.5% for maintenance. The mice were deemed dead when the following conditions occurred: Inability to move, no spontaneous breathing, and no heartbeat. Each tumor weight was documented. The animal protocol (approval no. EC-A-2022-5-23) was approved by the Research Ethics Committee of the 970th hospital of the PLA Joint Logistic Support Force (Yantai, China).

The data were analyzed and plotted as the mean ± SD using the Prism 7.0 software (Dotmatics). Significance of differences between two groups were compared with Student's t-test, paired t-test was used in the statistical analysis of clinical samples, and the other experiments adopted unpaired t-test to analyze the difference. For multi-group data comparisons, one-way ANOVA with Tukey's post hoc test was performed. The correlations between the expression of circBLNK, miR-188-3p, and GPX4 in OS tissues were analyzed using the Pearson's χ² test. The survival curves were established by the Kaplan-Meier plot (URL: http://kmplot.com/analysis/). P<0.05 was considered to indicate a statistically significant difference.

The circRNA expression data was downloaded from GEO (accession no. GSE140256) (31) and the differentially expressed circRNAs in OS were detected. CircBLNK expression was observed significantly elevated in OS tissues (n=3) relative to adjacent non-cancerous specimens (n=3) (Fig. 1A and B). Furthermore, circBLNK expression in the 40 pairs of OS and adjacent non-cancerous tissues was assessed by RT-qPCR. Consistently, circBLNK level in OS tissues was significantly higher than in control tissues (Fig. 1C). Furthermore, a positive association between circBLNK expression and tumor size (Fig. 1D) and cell differentiation (Fig. 1E) was detected. Based on the median expression value of circBLNK, the 40 patients were divided into circBLNK low and high expression groups. The Kaplan-Meier analysis exhibited that high circBLNK expression was associated with poor prognosis in OS patients (Fig. 1F). Collectively, the results of the present study revealed the anomalously high expression levels of circBLNK in OS tissues, which are associated with poor OS prognosis.

CircBLNK expression was further detected in four OS cell lines (HOS, SJSA-1, MG63 and U2OS) and a human normal osteoblast cell line (hFOB1.19). The four OS cell lines exhibited significantly higher circBLNK levels than hFOB1.19, with MG63 and U2OS cells exhibiting the highest (Fig. 2A). Afterwards, the subcellular localization of circBLNK in MG63 and U2OS cells was detected using the cell cytosolic/nuclear fractionation assay. The results revealed a specific cytoplasmic localization of circBLNK (Fig. 2B). Subsequently, the RNase R assay was used to evaluate the circular nature of circBLNK and it was revealed that RNase R efficiently digested liner BLNK but could not digest circBLNK (Fig. 2C). To dissect the role of circBLNK in OS progression, circBLNK expression in MG63 and U2OS cells was silenced by transfecting the cells with si-NC or si-circBLNK (Fig. 2D). Next, the CCK-8 and clonogenic assays were used to assess how circBLNK knockdown affected the cell proliferation ability and it was detected that the cell viability and colony-formation ability were suppressed (Fig. 2E and F). On the other hand, the cell apoptosis was enhanced when circBLNK expression was silenced (Fig. 2G). Collectively, the results of the present study revealed that circBLNK knockdown suppresses cell proliferation and enhances cell apoptosis.

CircInteractome revealed that circBLNK could bind to miR-188-3p (Fig. 3A). The luciferase reporter and RNA pull-down assays were then performed to check this possibility. Co-transfection of WT-circBLNK with miR-188-3p mimics into 293T cells diminished the luciferase reporter activity, whereas co-transfection of MUT-circBLNK with miR-188-3p mimics retained the normal activity compared with the co-transfection with mimic control (Fig. 3B). Furthermore, using RT-qPCR, miR-188-3p expression in MG63 and U2OS cells transfected with si-NC or si-circBLNK was determined. The results demonstrated that knockdown of circBLNK increased miR-188-3p expression (Fig. 3C). In parallel, miR-188-3p transcript abundance was determined in the 40 pairs of OS and adjacent non-cancerous tissues. The abundance of miR-188-3p in OS tissues was significantly lower than in adjacent non-cancerous specimens (Fig. 3D). In addition, circBLNK expression was detected to be negetively correlated with miR-188-3p expression in OS tissues (Fig. 3E). These analyses suggested that miR-188-3p expression is subjected to the regulation by circBLNK. Next, for the collection of direct evidence supporting that circBLNK targets miR-188-3p, the RNA pull-down assay was performed. The assay detected the enrichment of miR-188-3p in the circBLNK probe fraction compared with the NC probe fraction (Fig. 3F). Taken together, these findings indicated that circBLNK regulates the OS tumorigenesis by sponging miR-188-3p.

The miRDB revealed a high possibility that miR-188-3p bind to GPX4 (Fig. 4A). The luciferase reporter assay was then performed to validate this interaction. Co-transfection of miR-188-3p mimics and WT-GPX4 resulted in weaker luciferase reporter activity in 293T cells, whereas the cells co-transfected with miR-188-3p mimics and MUT-GPX4 showed normal luciferase reporter activity (Fig. 4B). Moreover, the RIP assay displayed that both miR-188-3p and GPX4 were effectively pulled down by Ago2, indicating the direct interaction between them (Fig. 4C and D). Afterwards, GPX4 expression was measured in the 40 pairs of OS and adjacent non-cancerous tissues by RT-qPCR. GPX4 expression was significantly elevated in OS tissues (Fig. 4E). Additionally, a positive correlation was detected between the abundance of circBLNK and GPX4, whereas a negative correlation was found between the abundance of miR-188-3p and GPX4 in OS tissues (Fig. 4F). Moreover, the levels of GPX4 transcripts and protein were all diminished in MG63 and U2OS cells transfected with si-circBLNK, which was reversed with the presence of miR-188-3p inhibitor (Fig. 4G and H). Furthermore, circBLNK knockdown suppressed the proliferation but enhanced the cell apoptosis of MG63 and U2OS cells, while these effects were reversed with the presence of miR-188-3p inhibitor (Fig. 4I and J). In summary, the results of the present study demonstrated that circBLNK promotes OS progression by regulating the miR-188-3p/GPX4 pathway.

Finally, the regulation of ferroptosis by the circBLNK/miR-188-3p/GPX4 signaling was investigated in OS cells. First, with the use of MTT assay, erastin, a ferroptosis inducer, was verified to activate cell death in MG63 and U2OS cells, whereas ferrostain-1, a ferroptosis inhibitor, could counteract the effect of erastin (Fig. 5A). Of note, ZVAD-FMK (apoptosis inhibitor) and necrosulfonamide (necroptosis inhibitor) did not affect the induction of cell death by erastin (Fig. 5A). Afterwards, using the MTT assay, the way circBLNK/miR-188-3p/GPX4 signaling affects ferroptosis induction by erastin was investigated. The percentage of cells undergoing cell death was revealed to be increased by circBLNK knockdown, an effect that was reversed by OE-GPX4-overexpressing cells (Fig. 5B). Since cell death is dependent on intracellular iron (Fe²⁺) level, it was then assessed whether the circBLNK/miR-188-3p/GPX4 signaling affects the intracellular Fe²⁺ level in MG63 andU2OS cells. The concentration of Fe²⁺ was demonstrated to be increased when circBLNK was knocked down, which was rescued by GPX4-overexpressing cells (Fig. 5C and D). Furthermore, circBLNK knockdown resulted in an enhancement of MDA and lipid ROS generation, which was reversed by GPX4-overexpressing cells (Fig. 5E and F). In the cells with circBLNK knockdown, the mitochondrial superoxide concentration was increased, while the mitochondrial membrane potential was diminished, which were eliminated by GPX4-overexpressing cells (Fig. 5G and H). These results suggested that circBLNK inhibits ferroptosis by regulating the miR-188-3p/GPX4 signaling in OS cells.

To confirm circBLNK is a bona fide tumor promoter in vivo, MG63 cells stably transfected with si-NC or si-circBLNK were used to inject nude mice subcutaneously. Tumor volume detection revealed a significant decrease in subcutaneous tumor volume in the si-circBLNK group vs. the si-NC group (Fig. S1A). Similarly, circBLNK knockdown significantly reduced the tumor weight (Fig. S1B). Collectively, the aforementioned data suggested that circBLNK promotes OS tumor growth in vivo.