Circular RNA circFIRRE drives osteosarcoma progression and metastasis through tumorigenic-angiogenic coupling

Abstract

Background: Disappointing clinical efficacy of standard treatment has been proven in refractory metastatic osteosarcoma, and the emerging anti-angiogenic regimens are still in the infantile stage. Thus, there is an urgent need to develop novel therapeutic approach for osteosarcoma lung metastasis.

Methods: circFIRRE was selected from RNA-sequencing of 4 matched osteosarcoma and adjacent samples. The expression of circFIRRE was verified in clinical osteosarcoma samples and cell lines via quantitative real-time polymerase chain reaction (RT-qPCR). The effect of circFIRRE was investigated in cell lines in vitro models, ex vivo models and in vivo xenograft tumor models, including proliferation, invasion, migration, metastasis and angiogenesis. Signaling regulatory mechanism was evaluated by RT-qPCR, Western blot, RNA pull-down and dual-luciferase reporter assays.

Results: In this article, a novel circular RNA, circFIRRE (hsa_circ_0001944) was screened out and identified from RNA-sequencing, and was upregulated in both osteosarcoma cell lines and tissues. Clinically, aberrantly upregulated circFIRRE portended higher metastatic risk and worse prognosis in osteosarcoma patients. Functionally, in vitro, ex vivo and in vivo experiments demonstrated that circFIRRE could drive primary osteosarcoma progression and lung metastasis by inducing both tumor cells and blood vessels, we call as "tumorigenic-angiogenic coupling". Mechanistically, upregulated circFIRRE was induced by transcription factor YY1, and partially boosted the mRNA and protein level of LUZP1 by sponging miR-486-3p and miR-1225-5p.

Conclusions: We identified circFIRRE as a master regulator in the tumorigenesis and angiogenesis of osteosarcoma, which could be purposed as a novel prognostic biomarker and therapeutic target for refractory osteosarcoma.

Keywords: Angiogenesis; LUZP1; Osteosarcoma; YY1; circFIRRE; miR-1225-5p; miR-486-3p.

Fig. 1

Identification and validation of upregulated circFIRRE in OS tissues and cells. A A Volcano plot illustrating the differential expression of circRNA in RNA-sequencing, where the red and blue points represent upregulated and downregulated circRNAs that met screening criteria as described. B Validation of differentially expressed circular RNAs by RT-qPCR (n=3 in each group). C The relative expression of circFIRRE in 104 paired clinical OS samples and normal controls were analyzed by RT-qPCR. D RNA–fluorescence in situ hybridization assays (FISH) were conducted in clinical OS and normal tissues to demonstrate circFIRRE expression using Cy3-labeled probes (red); DAPI-stained nuclei (blue). Scale bars=100 μm and 50 μm. E The relative expression of circFIRRE in human OS cell lines and normal osteoblasts (n=3 in each group). F-G RT-qPCR indicated the abundance of circFIRRE in cytoplasm of MG63 and U2OS cells. GAPDH and U6 were used as positive controls for cytoplasm and nucleus (n=3 in each group). H FISH assay demonstrated that circFIRRE was predominantly localized in the cytoplasm of MG63 and U2OS cells. Red indicates circFIRRE; DAPI-stained nuclei (blue). Scale bar=50 μm. I Schematic illustration presents the formation of circFIRRE. The trans-splicing site of circFIRRE validated by Sanger sequencing is highlighted by the red underline. J-K The relative expression of circFIRRE and linear FIRRE was analyzed by RT-qPCR at different time points after Actinomycin D treatment in MG63 and U2OS cells (n=3 at each time point). L-O The relative expression of circFIRRE and linear FIRRE was detected by RT-PCR and RT-qPCR in MG63 and U2OS cells in the presence of RNase R. P-Q The presence of circFIRRE was validated in MG63 and U2OS cells by RT-PCR. circFIRRE was amplified by divergent primers in cDNA, but not in genomic DNA. Values are presented as mean ± SD; the bar charts, line charts, error bars and dots represent the quantitative analysis of 3 independent experiments; (B, C, 2-tailed Student t test; E, one-way ANOVA; J, K, M and O, two-way ANOVA); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns = not significant

Fig. 2

Identification of circFIRRE as an independent risk factor for predicting OS prognosis. A-B Risk factors related to overall survival (OS) and disease-free survival (DFS) were identified by univariate and multivariate Cox regression analysis, and risk factors with significant difference are highlighted in red. C-D Forest plots of univariate and multivariate analysis for risk factors associated with OS and DFS. E Kaplan–Meier curves for OS, high circFIREE expression versus low circFIREE expression. F Kaplan–Meier curves for DFS, high circFIREE expression versus low circFIREE expression. G Multidimensional analysis of association between the circFIRRE expression level, clinical outcome (survival time, status) of osteosarcoma patients and risk score

Fig. 3

circFIRRE affects OS proliferation, migration and invasion in vitro. A Relative mRNA expression of circFIRRE and linear FIRRE in both MG63 and U2OS cells was examined by RT-qPCR after stable infection of either shRNAs lentivirus (sh-circFIRRE-1 and -2) or scramble shRNA lentivirus (sh-N.C.) (n=3 in each group). B-C CCK8 assay was applied to measure the cell viability influenced by circFIRRE knockdown at different time points in both MG63 and U2OS (n=6 at each time point). D-F EdU assay was performed to detect cell proliferation after circFIRRE silencing in MG63 and U2OS. S-phase entry was visualized by EdU incorporation (red); DAPI-stained nuclei (blue). Scar bar=200 μm. Image quantification conducted as described in methods (n=5 in each group). G-I Cell migration and invasion were detected by transwell assay after circFIRRE silencing in MG63 and U2OS cells. Scar bar=400 μm. Image quantification conducted as described in methods (n=5 in each group). J Relative expression of circFIRRE and linear FIRRE in both MG63 and U2OS cells was detected by RT-qPCR after transfection of circFIRRE overexpression vectors or scramble vectors (n=3 in each group). K-L CCK8 assay was applied to measure the cell viability after circFIRRE overexpression at different time points in MG63 and U2OS (n=6 at each time point). M-R EdU and Transwell assays were employed to detect cell proliferation, migration and invasion after circFIRRE overexpression in MG63 and U2OS. Scar bars=200 μl and 400 μm. Values are presented as mean ± SD; the bar charts, line charts, error bars and dots represent the quantitative analysis of 3 independent experiments; two-way ANOVA was used; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns = not significant

Fig. 4

circFIRRE can induce angiogenesis in vitro, ex vivo and in vivo. A Schematic description of the flow of magnetic bead cell sorting. B Validation of differentially expressed circFIRRE in primary endothelia cells by RT-qPCR (n=3 in each group). C Validation of differentially expressed circFIRRE in different endothelia cell lines by RT-qPCR (n=3 in each group). D HUVEC cells were transiently transfected with circFIRRE-siRNAs or scramble siRNA (si-N.C.). 48 hours after transfection, the relative circFIRRE expression in HUVEC cells were detected by RT-qPCR (n=3 in each group). E-F Tube formation assay was applied to determine cell tube formation ability after circFIRRE silencing in HUVEC cells. Scar bar=1mm. Image quantification were conducted as described in methods (n=3 in each group). G Representative immunofluorescence staining of aortic rings. BS1 lectin-FITC (green) indicated endothelial sprouts; α-SMA-Cy3 (red) stained supporting cells; DAPI-stained nuclei (blue). Scale bars=400 μm and 150 μm. H Representative images of CAM photographed on plastic dished after resection from eggs. Scale bars=4 cm and 2 cm. I-L Tube formation, aortic ring and CAM assays were conducted to examine angiogenesis level induced by circFIRRE overexpression, respectively. Values are presented as mean ± SD; the bar charts, error bars and dots represent the quantitative analysis of 3 independent experiments; (B, 2-tailed Student t test; C, D, one-way ANOVA; F, J, two-way ANOVA); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001

Fig. 5

YY1 activates circFIRRE transcription. A Wayne diagram showed the overlap between transcription factors of circFIRRE predicted by three different algorithms. B Relative mRNA expression of YY1 in 35 paired OS and adjacent normal samples were examined by RT-qPCR. C Pearson correlation analysis indicated that YY1 was positively correlated with circFIRRE in 35 paired samples. D The relative expression of YY1 was detected in chosen OS cell lines (MG63, U2OS) and control osteoblasts (n=3 in each group). E Schematic illustration of the binding motif of YY1 and the two possible YY1 binding sites in circFIRRE promoter. F Dual luciferase assay was performed to confirm the binding between YY1 and circFIRRE promoter, as well as two specific binding sites in HEK-293 T cells (n=3 in each group). G Three YY1-targeted siRNAs were transfected into MG63 and U2OS cells, and relative mRNA of YY1 were estimated by RT-qPCR (n=3 in each group). H-I The relative circFIRRE and linear FIRRE expression were detected in MG63 and U2OS cells after YY1 silencing by RT-qPCR (n=3 in each group). J-L The relative YY1, circFIRRE and linear FIRRE expression were detected in MG63 and U2OS cells after YY1 overexpression by RT-qPCR (n=3 in each group). Values are presented as mean ± SD; the bar charts, error bars and dots represent the quantitative analysis of 3 independent experiments; (B, F, 2-tailed Student t test; D, one-way ANOVA; G-L, two-way ANOVA); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns = not significant

Fig. 6

circFIRRE acts as a sponge for miR-486-3p and miR-1225-5p (A-C) Anti-AGO2 RNA immunoprecipitation (RIP) assay was performed to estimate AGO2 binding to circFIRRE in MG63 and U2OS cells, followed by RT-qPCR and RT-PCR detection, using IgG as the negative control (n=3 in each group). D Schematic illustration exhibited overlapping of the target miRNAs of circFIRRE predicted by five algorithms. E The relative expression of miR-486-3p and miR-1225-5p declined in 35 paired OS samples relative to adjacent normal controls through RT-qPCR detection, respectively. F-G Representative images of FISH were shown to illustrated the declined miR-486-3p and miR-1225-5p expression in OS samples compare with the adjacent normal samples. Scale bars=100 μm and 50 μm. H-I Pearson correlation analysis indicated a negative correlation between circFIRRE and two miRNAs in 35 paired samples. J-K The relative expression of miR-486-3p and miR-1225-5p was detected by RT-qPCR after transfecting circFIRRE siRNA or overexpression vectors in both MG63 and U2OS cells (n=3 in each group). L-M circFIRRE-specific biotin-labelled probe could successfully capture miR-486-3p and miR-1225-5p in MG63 and U2OS, using the oligo probe as the negative control (n=3 in each group). N-O Specific biotin-labelled miR-486-3p and miR-1225-5p probes could successfully capture circFIRRE relative to biotin-NC probe in both MG63 and U2OS (n=3 in each group). P Dual luciferase assay was performed to estimate the binding between circFIRRE and two miRNAs, after co-transfecting miRNA mimics and luciferase reporter plasmids into HEK-293 T (n=3 in each group). Q-R Typical fields of FISH were showed to illustrate the co-localization of circFIRRE and two miRNAs in both MG63 and U2OS. Scale bar=50 μm. Values are presented as mean ± SD; the bar charts, error bars and dots represent the quantitative analysis of 3 independent experiments; two-way ANOVA were used; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns = not significant

Fig. 7

circFIRRE promotes OS tumorigenesis and angiogenesis by miR-486-3p/miR-1225-5p-LUZP1 axis in vitro. A Schematic illustration exhibited overlapping of the target genes of miR-486-3p and miR-1225-5p predicted by three algorithms and RNA-sequencing data. B The relative expression of LUZP1 declined in 35 paired OS samples relative to adjacent normal controls through RT-qPCR detection, respectively. C-E Pearson correlation analysis illustrated that LUZP1 was positively correlated with circFIRRE, while negatively correlated with miR-486-3p and miR-1225-5p in 35 paired samples. F Dual luciferase assay was performed to verify whether miR-486-3p and miR-1225-5p could bind to predicted binding sites of LUZP1 (n=3 in each group). G-H Relative mRNA and protein expression of LUZP1 were suppressed under circFIRRE silencing, while the downregulation of LUZP1 was retarded by miR-486-3p and (or) miR-1225-5p inhibitors. I-L Transwell assay and tube formation assay were conducted in MG63, U2OS and HUVEC cells on the third day after co-transfection. Scar bars=400 μm and 1mm. Values are presented as mean ± SD; the bar charts, error bars and dots represent the quantitative analysis of 3 independent experiments; (B, G, 2-tailed Student t test; F, J, L, two-way ANOVA); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns = not significant

Fig. 8

circFIRRE promotes both primary OS progression and metastasis in vivo by sponging miRNAs. MG63 cells infected with empty virus (Group-A), sh-circFIRRE-1 (Group-B) or co-infected with sh-circFIRRE-1 and miR-486-3p+1225-5p sponges (Group-C) for in vivo assays. A-E Orthotopic xenograft tumor models (n=10 in each group). A IVIS imaging was applied to determine the OS lesions in situ. The representative images were acquired with the same exposure. B Micro-CT scans and 3-D reconstruction were performed to estimate bone destruction induced by tumorigenesis in situ. C-D Ki-67 and LUZP1 detection in bone lesions by IHC. E LUZP1 protein level in bone lesions was examined by western blot. F-N Tail vein metastasis models (n=10 in each group). F Representative IVIS imaging of metastatic tumor activity in the lung with the same exposure. G-I 3-D reconstructions of micro-CT. G The white arrowhead indicates the metastatic lesions in the lung. H-I Quantitation of tumor volumes and numbers. J-K Collected lungs and Hematoxylin and eosin (H&E) staining. K The black arrowhead indicates the metastatic lesions in the lung. L-N VEGF, CD31 and LUZP1 detection in lung lesions by IHC. O LUZP1 protein level in lung lesions was examined by western blot. P A proposed model of circFIRRE in modulating OS tumorigenesis, metastasis and angiogenesis via YY1-circFIRRE-miR-486-3p/miR-1225-5p-LUZP1 axis. Scale bars of IHC=100 μm and 50 μm