CircCDK14 Promotes Tumor Progression and Resists Ferroptosis in Glioma by Regulating PDGFRA

Abstract

CircRNAs have garnered significant interest in recent years due to their regulation in human tumorigenesis, yet, the function of most glioma-related circRNAs remains unclear. In this study, using RNA-Seq, we screened differentially regulated circRNAs in glioma, in comparison to non-tumor brain tissue. Loss- and gain-of-function strategies were used to assess the effect of circCDK14 on tumor progression both in vitro and in vivo. Luciferase reporter, RNA pull-down and fluorescence in situ hybridization assays were carried out to validate interactions between circCDK14 and miR-3938 as well as miR-3938 and PDGFRA. Transmission electron microscopic observation of mitochondria, iron and reactive oxygen species assays were employed for the detection of circCDK14 effect on glioma cells' sensitivity to erastin-induced ferroptosis (Fp). Our findings indicated that circCDK14 was overexpressed in glioma tissues and cell lines, and elevated levels of circCDK14 induced poor prognosis of glioma patients. CircCDK14 promotes the migration, invasion and proliferation of glioma cells in vitro as well as tumorigenesis in vivo. An evaluation of the underlying mechanism revealed that circCDK14 sponged miR-3938 to upregulate oncogenic gene PDGFRA expression. Moreover, we also found that circCDK14 reduced glioma cells' sensitivity to Fp by regulating PDGFRA expression. In conclusion, circCDK14 induces tumor in glioma and increases malignant tumor behavior via the miR-3938/PDGFRA axis. Hence, the miR-3938/PDGFRA axis may be an excellent candidate of anti-glioma therapy.

Keywords: CircCDK14; Ferroptosis; Glioma; PDGFRA; miR-3938.

Figure 1

Expression and characterization of circCDK14 in human glioma. (A) Clustered heatmap of significant differentially regulated circRNAs in glioma tissues and non-tumor brain tissues (|fold change >2, p<0.05). Red, up- and green, down-regulated. (B) Schematic illustration of genomic location and formation of circCDK14 (hsa_circ_0001721), derived from exons 3 circularization of CDK14 gene. The back-splice junction sequences of circCDK14 were confirmed by Sanger sequencing. Arrow, “head-to-tail” splice junction site. (C) PCR was analyzed to the circular RNA characterization of the circCDK14 (hsa_circ_0001721), using the divergent and convergent primers amplifying from the cDNA and gDNA of U251 cells, respectively. (D) qRT-PCR analysis of circCDK14 and linear CDK14 mRNA, in presence or absence of RNase R treatment for 30 min. (E) CircCDK14 levels, as quantified by qRT-PCR in human glioma cell lines (U87, U251 and SF126) and normal brain glial cells (HEB). circCDK14 expression level were significantly elevated in glioma cells, relative to HEB cells. (F) CircCDK14 levels in 76 glioma and 17 non-tumor brain tissues, and in 30 patients with I-II grade glioma and 46 patients with III-IV grade glioma. (G) Overall survival curve, based on circCDK14 levels, plotted with Kaplan-Meier methods and analyzed by rank test. (H) qRT-PCR analysis of circCDK14 expression location using nuclear and cytoplasmic fractions of U87 and U251 cells. U6 small nuclear RNA and GAPDH was endogenous control. Data expressed as mean±SD. *p< 0.05; **p< 0.01.

Figure 2

circCDK14 promotes the proliferation, migration and invasion of glioma. (A and B) Levels of circCDK14 and host gene CDK14 mRNA in U87 cells and U251 cells after transfection of circCDK14 overexpression plasmid or circCDK14 interference sequences, were detected by qRT-PCR. (C) Overexpressing circCDK14 promotes U87 cell proliferation, as determined by CCK-8. (D) CircCDK14 knockdown inhibits U251 cell proliferation, as determined by CCK-8. (E and F) Wound healing assay determined the migrating capacity of glioma cells after various treatments. (G and H) Transwell assay was used to detect the invasion activity of glioma cells following different treatments. Both scale bars are 100 µm. All experiments repeated thrice and averaged (mean±SD). Unpaired Students's t-test was employed for all data analyses. *p< 0.05; **p< 0.01; ***p< 0.001; ****p< 0.0001.

Figure 3

circCDK14 acts as sponge for miR-3938 in glioma. (A) Illustrations of the estimated docking sites of miRNA candidates to circCDK14. (B) Luciferase activity of pmirGLO-circCDK14-WT in cells after co-incorporation with miRNAs mimics. (C) Estimated docking sites of miR-3938 with circCDK14 was predicted. circCDK14 WT, circCDK14 wild-type luciferase reporter; circCDK14 MUT, circCDK14 mutantluciferase reporter. (D) U251 and U87 cells were co-transfected with pmirGLO -circCDK14-WT or pmirGLO -circCDK14-MUT and miR‐3938 mimics or negative control. Luciferase activity measured with luciferase reporter assays. (E) Colocalization between circCDK14 and miR-3938 were observed by RNA in situ hybridization (FISH) in U87 and U251 cells. DAPI staining was used for the nuclei. CircCDK14 probe was Cy3-labeled, and miR-3938 probe was FAM-labeled. (F) RNA pull-down assay was conducted to show the relative miR-3938 level in U87 and U251 cells lysates pulled down by circCDK14. (G) The expression of miR-3938 in 76 glioma and 17 non-tumor brain tissues were detected by qRT-PCR. (H) The overall survival curve, based on miR-3938 levels, was plotted with Kaplan-Meier methods and analyzed by rank test. All experiments repeated three times and data averaged and expressed as mean±SD.*p< 0.05; **p< 0.01.

Figure 4

circCDK14 regulates PDGFRA expression via miR-3938. (A) Putative binding sites of miR-3938 with PDGFRA was predicted. PDGFRA WT, PDGFRA wild-type pmirGLO luciferase reporter; PDGFRA MUT, PDGFRA mutant pmirGLO luciferase reporter. Wild-type and mutated 3'-UTR sequences of PDGFRA are shown. (B) U251 and U87 cells were co-transfected with pmirGLO-PDGFRA -WT or pmirGLO -PDGFRA -MUT and miR‐3938 mimics or negative control. Luciferase activity was measured with luciferase reporter assays. (C) RNA pull-down assay was conducted to show the relative miR-3938 level in U87 and U251 cells lysates pulled down by PDGFRA (D) PDGFRA expressed was affected by miR-3938 levels. PDGFRA mRNA levels, as evidenced by qRT-PCR in U87 cells incorporated with miR-3938 mimics and U251 cells incorporated with miR-3938 inhibitors. (E and F) U87 cells were incorporated with circCDK14 plasmid or co-transfected with circCDK14 plasmid and miR-3938 mimics. RT-PCR and Western blot detected PDGFRA mRNA and protein expressions (G and H) U251 cells were incorporated with circCDK14 interference sequences or co-transfected with circCDK14 interference sequences and miR-3938 inhibitors. RT-PCR and Western blot detected PDGFRA mRNA and protein levels. All experiments repeated three times and data averaged and expressed as mean±SD. *p< 0.05, **p< 0.01, ***p< 0.001.

Figure 5

Downregulated PDGFRA turnovers circCDK14 induced malignant phenotype. (A) qRT-PCR detects the mRNA PDGFRA levels after transfection of si-PDGFRA. (B) Cell proliferation was measured by CCK8 in U87 cells following different treatments. (C) Wound healing assay detected U87 cell migrating capacity after various treatments. (D) Transwell assay detected the invasion activity of U87 cells following various treatments. All experiments repeated three times and data averaged and expressed as mean±SD. *p< 0.05, ***p< 0.001.

Figure 6

PDGFRA expression is upregulated in glioma and negatively associated with ferroptosis. (A) Pan-cancer PDGFRA levels between TCGA-derived tumor tissues and TCGA- and GTEx-derived normal tissues. Asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, independent variable t-test) represents marked alterations in median PDGFRA mRNA levels between tumors and normal control. (B) GSEA indicated significant enrichment of Fp phenotype in the low PDGFRA expression patients. (C) The Spearman's correlations between PDGFRA expression and the FPI. TCGA, The Cancer Genome Atlas; GTEx, Genotype-Tissue Expression; GSEA, Gene Set Enrichment Analysis; FPI, Fp potential index. (D) Transmission electron microscope was used to observe mitochondria in U251 cells with different treatment. Scale bar: 5 µm and 2 µm. In the lower set of figures, arrows point to the mitochondria. The small box in the upper left corner shows the enlarged mitochondria. Mock, control group; NC+Erastin, U251 cells transfected by negative control sequence following 10 µM erastin treatment for 24 h; siPDGFRA +Erastin, U251 cells transfected by PDGFRA interference sequence1 following 10 µM erastin treatment for 24 h.

Figure 7

circCDK14 inhibits glioma cells' sensitivity to Fp. (A) Transmission electron microscope was used to observe mitochondria in U251 cells with different treatments. Scale bar: 5 µm and 2 µm. In the lower set of figures, arrows point to the mitochondria. The small box in the upper left corner shows the enlarged mitochondria. Mock, control group; NC+Erastin, U251 cells transfected by negative control sequence following 10 µM erastin treatment for 24 h; si-circRNA1+Erastin, U251 cells transfected by circCDK14 interference sequence1 following 10 µM erastin treatment for 24 h. (B) GPX4, SLC7A11 and NRF2 protein expressions in U251 cells or circCDK14 knockdown U251 cells were detected by western blot. (C) Fe²⁺ level was measured in U251 cells or circCDK14 knockdown U251 cells treatment with 10 µM erastin for 24 h by spectrophotometric method. (D) ROS levels were detected in U251 cells or circCDK14 knockdown U251 cells treated with 10 µM erastin for 24 h by H2DCFH-DA fluorescent probe assay. (E) The protein levels of PDGFRA, p-PDGFRα, GPX4, SLC7A11 and NRF2 in U87 cells with different treatments, as evidenced by Western blot. Crenolanib, PDGFRα signaling inhibitor. (F) The protein levels of p-PDGFRA, ZEB1, vimentin and E-cadherin in U251 cells or circCDK14 knockdown U251 cells, as evidenced by western blot. (G) PDGFRA, p-PDGFRα, ZEB1, vimentin and E-cadherin protein levels in U87 cells with different treatments were detected by Western blot. Crenolanib, PDGFRα signaling inhibitor. (H) Schematic illustration showing the biological function and mechanism of circCDK14 in glioma. All experiments repeated three times and data averaged and expressed as mean±SD. *p< 0.05, ***p< 0.001, ****p< 0.0001.

Figure 8

CircCDK14 promotes the growth of gliomas in vivo. (A) qRT‐PCR analyses for the expression of circCDK14 after stable transfection sh-circCDK14 and sh-NC in U251 cells. (B) The volume of subcutaneous tumor was measured every seven days. Formula: V(mm³) = length× width²/2. (C) Photographs of subcutaneous glioma tumors in nude mice (n=5 each group). (D) Tumor weight was markedly reduced in sh-circCDK14 group, relative to controls. (E and F) qRT-PCR detects the circCDK14 and miR-3938 levels in nude mice tumors from different groups. (G) Western blotting demonstrated the protein levels of PDGFRA, p-PDGFRα, ZEB1, Vimentin, E-cadherin, SLC7A11 and GPX4 in nude mice tumors from different groups. All experiments repeated three times and data averaged and expressed as mean±SD. *p< 0.05; ***p< 0.001; ****p< 0.0001.