Circular RNA LPAR3 enhances GPX4 expression by targeting miR-196a-5p to suppress ferroptosis of osteosarcoma cells

CircLPAR3 inhibition represses osteosarcoma cell viability and invasion/migration but enhances apoptosis

To identify the potential circular RNAs which are involved in the modulation of osteosarcoma progression, we performed the circular RNA microarray analysis in clinical osteosarcoma samples and primary chondroma samples. Among the differently expressed circular RNAs, we identified that circLPAR3 presented a most significant elevation in the osteosarcoma samples (Figure 1A). Our data validated that the expression of circLPAR3 was enhanced in the clinical osteosarcoma tissues (n=50) and osteosarcoma cell lines (Figure 1B, 1C). We treated the U2OS and MG63 cells with LPAR3 shRNA and found that circLPAR3 expression remarkably down-regulated in the cells (Figure 1D). Functionally, the cell proliferation was repressed by circLPAR3 knockdown in the U2OS and MG63 cells (Figure 1E). Conversely, apoptosis was enhanced by circLPAR3 shRNA in the cells (Figure 1F, 1G). Impressively, the depletion of circLPAR3 attenuated the invasion/migration of U2OS and MG63 cell lines (Figure 1H, 1I).

Figure 1. CircLPAR3 inhibition represses proliferation, invasion and migration but enhances apoptosis of osteosarcoma cells. (A) Heat map and hierarchical clustering analysis of circRNAs which were differentially expressed between clinical osteosarcoma samples (n=3) and primary chondroma samples (n=3) in circular RNA microarray analysis. Each column represented the expression profile of a tissue sample), and each row corresponds to a circRNA (log foldchange (FC) ≥ 2 and P < 0.05). Red color represented up-regulated circRNAs, and green color represented down-regulated circRNAs. (B) The expression of circLPAR3 was detected in clinical osteosarcoma samples (n=50) by qPCR. (C) The expression of circLPAR3 was determined by qPCR in hFOB1.19 cell line and osteosarcoma cell lines, including HOS, U2OS, 143B, MG63, and SJSA1. (D–I) The functional analysis was performed in the U2OS and MG63 cells treated with control shRNA (shcontrol) or circLPAR3 shRNA. (D) The expression of circLPAR3 was measured by qPCR in the cells. (E) The cell viability of U2OS and MG63 cells was analyzed by MTT assays. (F, G) The apoptosis of U2OS and MG63 cells was assessed by flow cytometry analysis. (H, I) The invasion and migration of U2OS and MG63 cells were detected by transwell assays. ** P < 0.01.

CircLPAR3 knockdown promotes osteosarcoma cells ferroptosis

We then were interested in the function of circLPAR3 in ferroptosis of osteosarcoma cells. The treatment of ferroptosis stimulator termed erastin reduced the cell proliferation relative to DMSO, and circLPAR3 shRNA further reinforced this effect (Figure 2A, 2B). But we failed to observe the significant effect of circLPAR3 shRNA on the ferroptosis inhibitor called ferrostatin-1 (Figure 2A, 2B). In addition, the levels of iron, Fe²⁺, and ROS were significantly increased by circLPAR3 suppression in osteosarcoma cells (Figure 2C–2E).

Figure 2. CircLPAR3 knockdown promotes ferroptosis of osteosarcoma cells. (A, B) The cell viability was analyzed by MTT assays in U2OS and MG63 cells co-treated with circLPAR3 shRNA and erastin (5 mmol/L) or ferrostatin (1 mmol/L). (C–E) The analysis was performed in the U2OS and MG63 cells treated with circLPAR3 shRNA. The levels of iron (C), Fe²⁺ (D), and lipid ROS (E) were analyzed by the corresponding measurement kits. ** P < 0.01.

GPX4 participates in circLPAR3-regulated ferroptosis in osteosarcoma cells

We then explored the potential mechanisms of circLPAR3-mediated ferroptosis. Given that SLC7A11 and GPX4 are the essential regulators and the molecular markers of ferroptosis, we were interested in whether circLPAR3 modulates SLC7A11 and GPX4 expression in the regulation of ferroptosis of osteosarcoma. Significantly, we found that the protein expression of SLC7A11 and GPX4 was repressed by circLPAR3 shRNA, in which the re-introduction of GPX4 rescued the expression of these factors in osteosarcoma cells (Figure 3A). Moreover, the accumulation of iron, Fe²⁺, and ROS was induced by circLPAR3 depletion, and the overexpression of GPX4 could attenuate this phenotype in osteosarcoma cells (Figure 3B–3D). We then validated that SLC7A11 and GPX4 were enhanced in clinical osteosarcoma tissues (n=50) (Figure 3E, 3F). We then selected GPX4 for the further analysis.

Figure 3. GPX4 participates in CircLPAR3-regulated ferroptosis in osteosarcoma cells. (A–D) The analysis was performed in the U2OS and MG63 cells treated with circLPAR3 shRNA and GPX4 reconstituted plasmid. (A) Western blot detected the GPX4 and SLC7A11 expression. The levels of iron (B), Fe²⁺ (C), and lipid ROS (D) were analyzed by the corresponding measurement kits. (E) The expression of SLC7A11 was measured in clinical osteosarcoma samples (n=50) by qPCR. (F) The expression of GPX4 was detected in clinical osteosarcoma samples (n=50) by qPCR. ** P < 0.01.

CircLPAR3 activates GPX4 expression by targeting miR-196a-5p

To determine the potential mechanism of circLPAR3 modulating GPX4 expression, we evaluated the potential interacted miRNAs with circLPAR3 and GPX4 using ENCORI database. In the potential miRNAs that both interacted with circLPAR3 and GPX4, RNA-pull down analysis identified that miR-196a-5p presented the highest interaction enrichment with circLPAR3 in U2OS cells (Figure 4A). We confirmed the interaction of miR-196a-5p with circLPAR3 and GPX4 mRNA 3’UTR (Figure 4B). The miR-196a-5p mimic notably enhanced its expression in the osteosarcoma cells (Figure 4C). MiR-196a-5p attenuated luciferase activities of both circLPAR3 and GPX4 mRNA 3’UTR in osteosarcoma cells (Figure 4D, 4E). The circLPAR3 shRNA induced miR-196a-5p expression and miR-196a-5p mimic suppressed GPX4 expression in osteosarcoma cells (Figure 4F, 4G). Significantly, the GPX4 levels were alleviated followed by circLPAR3 inhibition but miR-196a-5p inhibitor could reverse this effect in the U2OS and MG63 cells (Figure 4H). Then, we validated that miR-196a-5p expression was reduced in clinical osteosarcoma tissues (n=50) and negatively correlated with GPX4 expression in the tissues (Figure 4I, 4J).

Figure 4. CircLPAR3 activates GPX4 expression by targeting miR-196a-5p. (A) The interaction of circLPAR3 with miRNAs was analyzed by RNA-pull down assays in U2OS cells. (B) The prediction analysis using ENCORI database assessed the interaction of miR-196a-5p with circLPAR3 and GPX4 mRNA 3’UTR. (C–E) The analysis was performed in the U2OS and MG63 cells treated with miR-196a-5p mimic. (C) The qPCR measured the miR-196a-5p expression. (D, E) The luciferase reporter gene assays analyzed luciferase activities of circLPAR3 and GPX4 mRNA 3’UTR. (F) The qPCR measured the miR-196a-5p expression in the U2OS and MG63 cells treated with circLPAR3 shRNA. (G) Western blot detected the GPX4 expression in the U2OS and MG63 cells treated with miR-196a-5p mimic. (H) Western blot tested GPX4 expression in the U2OS and MG63 cells co-treated with circLPAR3 shRNA and miR-196a-5p inhibitor. (I) The expression of miR-196a-5p was detected in clinical osteosarcoma samples (n=50) by qPCR. (J) The correlation of miR-196a-5p expression with GPX4 expression was analyzed by qPCR in clinical osteosarcoma samples (n=50). * P < 0.05, ** P < 0.01.

MiR-196a-5p contributes to osteosarcoma cell ferroptosis involving GPX4

Furthermore, we detected influence of miR-196a-5p/GPX4 on ferroptosis of osteosarcoma cells. Our data demonstrated that miR-196a-5p mimic reinforced the inhibitory effect of erastin on U2OS and MG63 cell proliferation, which was rescued by GPX4 overexpression (Figure 5A, 5B). Moreover, the accumulation of iron, Fe²⁺, and ROS was induced by miR-196a-5p mimic depletion, and the re-introduction of GPX4 could attenuate this phenotype in osteosarcoma cells (Figure 5C–5E).

Figure 5. MiR-196a-5p contributes to osteosarcoma cell ferroptosis involving GPX4. (A, B) MTT assays analyzed cell viability in the U2OS and MG63 cells co-treated with miR-196a-5p mimic and erastin (5 mmol/L) or DMSO. (C–E) The analysis was performed in the U2OS and MG63 cells treated with miR-196a-5p mimic. The levels of iron (C), Fe²⁺ (D), and lipid ROS (E) were analyzed by the corresponding measurement kits. ** P < 0.01.

CircLPAR3/miR-196a-5p/GPX4 axis is involved in regulating tumor growth in vivo

We then observed that the inhibition of circLPAR3 attenuated the tumor size, tumor weight, and tumor volume in the tumorigenicity analysis of nude mice (Figure 6A–6C). Moreover, miR-196a-5p expression was induced but GPX4 levels were reduced in the circLPAR3 knockdown group (Figure 6D, 6E). The levels of lipid ROS and iron were induced by circLPAR3 depletion in the mice (Figure 6F, 6G).

Figure 6. CircLPAR3/miR-196a-5p/GPX4 axis is involved in regulating tumor growth in vivo. (A–E) Tumorigenicity assay in the nude mice (n = 5) analyzed the tumor growth of U2OS cells treated with control shRNA (shcontrol) or circLPAR3 shRNA. The tumor images (A), tumor volume (B), and weight (C) were presented. (D) The qPCR detected miR-196a-5p expression in the tumor tissues. (E) Western blot analysis determined GPX4 expression in the tumor tissues. The levels of lipid ROS (F) and iron (G) were analyzed by the corresponding measurement kits. ** P < 0.01.

We then identified that miR-196a-5p attenuated the tumor size/weight/volume in the tumorigenicity analysis of nude mice (Supplementary Figure 1A–1C). The enhanced miR-196a-5p expression was validated in the tumor tissues of the mice (Supplementary Figure 1D). Meanwhile, the GPX4 levels were inhibited by miR-196a-5p mimic in the tumor tissues of the mice (Supplementary Figure 1E).

Cell culture

The hFOB1.19 and osteosarcoma 143B, HOS, SJSA-1, U2OS, and MG63, cell lines were bought (American Type Tissue Culture Collection, ATCC). The cells were cultured at a condition of 37° C with 5% CO₂ in DMEM (BI, USA) with fetal bovine serum (10%, BI, USA), penicillin (100 units/mL, BI, USA), streptomycin (0.1 mg/mL, BI, USA). The miR-196a-5p mimic/inhibitor, circLPAR3 shRNA, and pcDNA3.1-GPX4 vector were obtained (QingKe, China).

Osteosarcoma samples

Primary chondroma and osteosarcoma tissues were collected from 50 patients before the neo-adjuvant chemotherapy. The patients were histologically characterized by a pathologist using the criteria (World Health Organization). The application of the samples was underwritten informed consent of the patients, were approved by the Ethics Committee of Linyi People's Hospital.

Circular RNA microarray analysis

The circular RNA microarray analysis was performed to explore the potential correlation of circular RNAs with osteosarcoma (BGI, China). Total RNA was obtained from 3 chondroma and osteosarcoma tissues, and linear RNAs were removed by RNase R digestion. After that, the enriched circRNAs were augmented and transcribed in fluorescent cRNAs, and analyzed by Arraystar Human circRNA Array (8 × 15K; Arraystar). The Agilent scanner G2505C scanned the microarray, the collected images were analyzed through the Agilent feature extraction software (11.0.0.1 version), and the obtained data were analyzed.

Apoptosis analysis

To detect apoptosis, U2OS and MG63 cells were treated with FITC-Annexin V/PI apoptosis kit (Beyotime). Briefly, cells were treated with FITC-Annexin V and PI (5 μl) for 30 minutes in dark, resuspended in PBS and subjected to flow cytometry (BD Bioscience).

MTT assays

Cells were seeded in a 96-well plate (2 × 10⁴/well) after transfection with indicated vectors and/or oligonucleotides. The cells were then exposed to MTT reagent (5 mg/mL) for four hours in the standard culture incubator. The absorbance values at 570nm were measured by a micro plate reader (Bio-Rad, USA).

Luciferase reporter gene assay

The wild-type (WT) or mutant (MUT) circLPAR3 or 3’UTR of GPX4 sequences were synthesized and inserted into pmirGLO vectors. Cells were plated in a 24-well plate, and were transfected with miR-196a-5p mimics or negative control, along with the WT or MUT vectors for 48 hours. A Dual-Luciferase Reporter Assay System (Promega, USA) was adopted to detect the luciferase activity.

ROS and iron levels measurement

The reactive oxygen species (ROS) and iron analysis were performed as the previous report [34]. The ROS levels were detected by DCFH-DA staining. The levels of iron and Fe2+ were analyzed by Iron Assay Kit in the cells.

Transwell assays

The invasion/migration of U2OS and MG63 cells was analyzed by using transwell chambers (Corning, USA). U2OS and MG63 cells were plated in the upper chamber with FBS-free DMEM medium. The lower chambers were filled with DMEM for 48 hours. The upper chamber was treated with paraformaldehyde (4%), dyed by crystal violet (0.1%) for 20 minutes, followed by the photograph and counting using a microscope. To detect cell invasion, the upper chambers were covered by Matrigel (BD Bioscience, USA).

RNA pulldown

Biotin-labeled circLPAR3 were synthesized by QIAGENE, and transfected to U2OS cells for 48 hours. The cells were then lysed and hatched with magnetic beads (Thermo) for 3 hours, followed by washing and detection by real-time PCR to measure the level of miRNAs.

Western blot analysis

Mice tissues or cells were homogenized by RIPA lysis buffer (Thermo). Proteins (30μg) were divided on SDS-polyacrylamide gel (SDS-PAGE), blotted onto polyvinylidene difluoride (PVDF) membranes. After that, the blots were blocked by 5% skim milk in PBST, incubated with specific primary antibodies against GPX4 (1:1000, Abcam, USA), SLC7A11 (1:1000, Abcam, USA), and β-actin (1:1000, Abcam, USA) at 4° C overnight. The blots were then hatched with secondary anti-rabbit or anti-mouse IgG conjugated with horse radish peroxidase (HRP) (Abcam, USA) at room temperature for 1 hour. The blots were then infiltrated with ECL (Thermo) and visualized in a Gel image system.

Quantitative reverse transcription-PCR (qRT-PCR)

The total RNA was extracted with Trizol reagent (Thermo) following a standard protocol. The cDNA was synthesized by using a PrimeScript™ 1st Strand cDNA Synthesis Kit (TaKaRa, Dalian, China). Quantitative real time PCR was conducted with Power Up SYBR Green Master Mix (Applied Biosystem, USA). Data were calculated by the 2^-ΔΔCt methods with GAPDH and U6 as reference genes, respectively. Primers were listed: GPX4: sense, 5′-GAGGCAAGACCGAAGTAAACTAC -3′, antisense, 5′-CCGAACTGGTTACACGGGAA -3′; miR-196a-5p: sense, 5′-CCGACGTAGGTAGTTTCATGTT -3′, antisense, 5′-GTGCAGGGTCCGAGGTATTC-3′; circLPAR3: sense, 5′-GAGTTACCTTGTTTTCTGGACA-3′, antisense: 5′-TGGAGAAGTGAACATCCTAAG-3′; GAPDH: sense, 5′-CATGGGTGTGAACCATGAGA-3′, antisense, 5′-CAGTGATGGCATGGACTG-TG-3′; U6: sense, 5′-CGCGCTTCGGCAGCACATATACT-3′, antisense, 5′-ACGCTTCACGAATTTGCGTGTC-3′.

Analysis of tumorigenicity in nude mice

ALL animal experiments in this work were conducted under the approval of Ethics Committee of the the Animal Ethics Committee of Linyi People's Hospital. Balb/c nude mice (4-weeks old) were ordered from Vital River Laboratory (China) and fed in a SPF environment. The mice were randomly divided into two groups, then U2OS cells (1×10⁷/mouse) were hypodermically injected in the nude mice. Tumor volume was analyzed by width (mm)² × length (mm)/2.

Statistical analysis

Data were shown by means ± standard deviation. Student’s t test or one-way analysis of variance with Dunnett’s multiple comparisons test were applied for comparison between two or multiple groups. P < 0.05 were statistically significant.