Circular RNA PRKCI promotes glioma cell progression by inhibiting microRNA-545

We here tested expression and potential functions of circular RNA PRKCI (circPRKCI) in human glioma. Our results show that circPRKCI is upregulated in human glioma tissues and glioma cells, correlating with downregulation of its potential target, microRNA-545 (miR-545). In A172 and primary human glioma cells, shRNA-mediated silencing of circPRKCI inhibited cancer cell growth, survival, proliferation, and migration. Conversely, ectopic circPRKCI overexpression promoted A172 cell progression. miR-545 is the primary target of circPRKCI in glioma cells. Forced overexpression of miR-545 mimicked circPRKCI shRNA-induced actions, inhibiting glioma cell survival and proliferation. In contrast, miR-545 inhibition, by a lentiviral antagomiR-545 construct, reversed circPRKCI shRNA-induced anti-A172 cell activity. Importantly, neither circPRKCI shRNA nor circPRKCI overexpression was effective in miR-545-knockout (Cas9 method) A172 cells. Importantly, the subcutaneous and orthotopic A172 xenograft growth was significantly inhibited by circPRKCI silencing. Collectively, circPRKCI promotes human glioma cell progression possibly by inhibiting miR-545. Targeting circPRKCI-miR-545 cascade could efficiently inhibit human glioma cells.


Introduction
Glioma, the most common brain tumor and a global health threat, causes significant mortalities each year [1][2][3] . Current therapeutic options for glioma have failed to substantially improve patients' prognosis [4][5][6] . High-grade (grade III-IV) glioma (i.e., glioblastoma) has one of the worst survival among all malignancies [1][2][3] , which is possibly due to its molecular heterogeneity 1 . It is therefore urgent to further explore the pathological mechanisms for glioma progression, which could help to develop novel and more efficient molecularly-targeted therapies [7][8][9] . microRNA (miRNAs), long non-coding RNAs (LncRNAs), and circular RNAs (circRNAs) are noncoding RNAs 10,11 . Recent studies have proposed the pivotal roles of these non-coding RNAs in glioma progression 12 . Unlike other non-coding RNAs, circRNAs can form highly stable circular structure via joining of the 3′ and 5′ terminals 11,13,14 . circRNAs function as miRNAs sponges, regulating gene expression in cancer cells 11,14,15 . Dysregulation of circRNAs is often detected in glioma 12 . Qiu et al. have recently discovered a novel and oncogenic circRNA, namely circRNAs PRKCI (circPRKCI) 16 . circPRKCI, with 1,484-bp long, is back-spliced of two exons (15 and 16) of PRKCI gene located at 3q26.2 16 . circPRKCI is upregulated in lung adenocarcinoma in part due to the amplification of 3q26.2 locus, promoting cancer cell proliferation and tumorigenesis 16 . circPRKCI is mainly present in the cytoplasm, sponging miR-545 and miR-589, thereby abolishing the suppressing of their target, the transcription factor E2F7 16 . Shi et al. have shown that circPRKCI promotes AKT3 expression and esophageal squamous cell proliferation by sponging miR-3680-3p 17 . In human hepatocellular carcinoma (HCC) cells circ-PRKCI functions as the sponge of miR-545, promoting cell survival 18 .The expression and potential functions of circPRKCI in human glioma cells are tested in the present study.

Chemicals and reagents
The antibodies utilized in this study were purchased from Abcam (Cambridge, MA). The reagents for cell culture were purchased from Hyclone (Logan, UT). Puromycin, polybrene and all other chemicals were provided by Sigma-Aldrich (St. Louis, Mo). All primers, sequences, virus, and expression constructs were designed, provided, and sequence verified by Shanghai Genechem Co. (Shanghai, China).

Cell culture
The established A172 glioma cells were provided by Dr. Cao 19 . Brain cortical tissues were obtained from 15-weekold normal fetal brains (prepared from Dr. Zhang 20 ), dissociated, digested, and filtered as previously described 21,22 . The dissociated cells were centrifuged and resuspended in MEM medium with applied supplements 22 . Cells were seeded at a density of 2 × 10 6 cells/mL on poly-lysine-coated tissue culture flasks. For neuronal culture, astrocytes were limited by 4-day treatment with FDU (Sigma). Neuronal cultures were grown for eight days (day in vitro 8, DIV8) before any further experiments. The primary human astrocytes (95% positive for glial fibrillary acidic protein) were provided from Dr. Cao 19 , prepared from 12-week cortical tissues from normal fetal brains. All procedures were approved by the Ethics Committee of Wenzhou Medical University.

Primary human glioma cells and tissues
In this study the primary human glioma cells were provided by Dr. Cao 19,23 , derived from three writteninformed consent glioma patients, named as "Pri-1/-2/-3". Primary human glioma cells were cultured in complete RPMI medium with necessary supplements and antibiotics 24,25 . The human glioma tissues and paired paracancer normal brain tissues were again provided by Dr. Cao 19,23 , stored in liquid nitrogen and subjected to further biomedical analyses. The protocols of this study were approved by the Ethics Committee of Wenzhou Medical University, in according to Declaration of Helsinki.

Quantitative real-time PCR (qPCR)
Total cellular and tissue RNAs were isolated by the Trizol reagent (Invitrogen, Grand Island, NY) and quantified. mRNA expression was tested by using the SYBR GREEN PCR Master Mix (Applied Biosystems, Beijing, China) under an ABI 7600 fast Real-time PCR System (Applied Biosystems). Relative expression of targeted mRNAs was calculated by 2 −ΔΔCt method, using GAPDH as the internal control. circPRKCI and miR-545 levels were tested by the TransStartTM SYBR Green qPCR Supermix (TransGen Biotech, Beijing, China), using U6 small nuclear RNA as the internal control. All the primers were listed in Table. 1.

Western blotting
Equivalent amounts of total cellular lysates (40 μg per treatment) were separated by 10-12% of SDS-PAGE gels, then transferred to the polyvinylidene fluoride (PVDF) blots (Merck Millipore, Darmstadt, Germany). After blocking in 10% non-fat milk, the blots were incubated with the applied primary antibodies, followed by incubation with corresponding secondary antibodies. Antibodyantigen binding was detected by an enhanced chemiluminescence (ECL) substrate kit (Invitrogen), with the results quantified by an ImageJ software (NIH, Bethesda, MD).
Colony formation assay A172 cells were initially seeded at 1 × 10 4 cells per 10cm dish. Colony formation assays were conducted for 10 days, and the colonies were fixed and stained (with 1% crystal violet solution). The number of colonies was counted manually.
In vitro migration assay A172 cells (1 × 10 5 cells in 300 μL serum-free medium) were seeded into the upper part of each "Transwell" chambers (12-μm pore size, BD Biosciences, Heidelberg, Germany). The lower compartments were filled with medium with 10% FBS. Following incubation for 24 h, nonmigrated cells on the upper surface were wiped out. The migrated cells, on the lower surface, were fixed and stained.

EdU assay of proliferation
Cells were seeded into six-well plates at 6 × 10 4 cells per well, and cultured for 48 h. An EdU (5-ethynyl-20-deoxyuridine) Apollo-567 Kit (RiboBio, Guangzhou, China) was applied. EdU and DAPI dyes were added to glioma cells for additional 4 h. Under a fluorescent microscope cell nuclei were visualized. For each condition total 300 nuclei in five random views were included to calculate EdU ratios (EdU/DAPI × 100%).

Annexin V FACS assay
Following the applied genetic treatments, Annexin V-FITC and Propidium Iodide (PI) dyes (each at 10 μg/mL, BD Pharmingen, San Diego, CA) were added for 30 min under the dark at room temperature. Cell apoptosis was analyzed by a flow cytometry machine (Beckman Coulter, Brea, CA).

miR-545 inhibition
The miR-545 inhibitor precursor, provided by Shanghai Genechem, was sub-cloned into the GV248 vector, cotransfected into 293 T cells with lentivirus package plasmids to generate antagomiR-545 lentivirus (LV-antag-omiR-545). The latter was transduced to A172 cells for 24 h. Puromycin was added again to select the stable cells. miR-545 inhibition in the stable cells was confirmed by qPCR.

miR-545 knockout
The CRISPR/Cas9 miR-545-KO lentivirus, with sgRNA targeting the pri-miR-545 sequence, was synthesized and verified by Shanghai Genechem Co. (Shanghai, China), added to A172 cells. The infected cells were than cultured in puromycin selection medium until stable cells were achieved. Cells were further transfected with or without the lentiviral circPRKCI shRNA construct ("LV-cir-cPRKCI shRNA") or the lentiviral circPRKCI overexpression construct ("LV-circPRKCI"). Expression of miR-545 and circPRKCI was tested by qPCR.

RNA immunoprecipitation (RIP)
A172 cells were lysed in complete RIP lysis buffer (purchased from Beyotime Biotechnology, Suzhou, China), and the cell extracts (800 μg lysates per treatment) were incubated with the magnetic beads conjugated with anti-Argonaute 2 (Ago2, Sigma) or control anti-IgG antibody (Sigma) for 12 h. The beads were washed and incubated with Proteinase K. Finally, the purified RNA was subjected to qPCR analysis, with results normalized to the Input control.

A172 xenograft assay
The severe combined immunodeficient (SCID) mice (4-5 week old, about 18 g weight) were provided by the Animal Center of Suzhou University (Suzhou, China), inoculated via subcutaneous (s.c.) injection with five million A172 cells with circPRKCI shRNA or control shRNA, in 200 µL of Matrigel gel/basal medium. Mouse body weights and tumor volumes were measured weekly, and tumor volume calculated by a described formula 19 . For intracranial tumor implantation, A172 glioma cells (1 × 10 6 cells) were implanted as described 27 . All animal procedures were approved by the Ethics Committee of Wenzhou Medical University.

Statistics
Statistical analyses were performed by SPSS 21.0 software (SPSS Inc., Chicago, IL). All data were presented as mean ± standard deviation (SD). Statistical differences were performed by one-way ANOVA within multiple comparisons with post hoc Bonferroni test. To determine significance between two treatment groups, the two-tailed t-tests were carried out. P values < 0.05 were considered statistically significant.

circPRKCI is upregulated in human glioma tissues and cells
First, circPRKCI expression in human glioma tissues was examined. A total of five pairs of fresh glioma tissues ("T", from Dr. Cao 19 ) and parecancer normal brain tissues ("N") were obtained. qPCR assays were performed to test circPRKCI expression. Results, in Fig. 1a, demonstrated that circPRKCI levels are significantly upregulated in all tested glioma tissues, when compared its levels in the normal brain tissues. Furthermore, circPRKCI is upregulated in A172 glioma cells and in the primary human glioma cells ("Pri-1/-2/-3", see Methods) ( Fig. 1b). While its levels are low in primary human neuronal cultures and human astrocytes (Dr. Cao 19 ) (Fig. 1b).
It has been previously shown that circPRKCI functions as the sponge of miR-545, a tumor-suppressive miRNA 16 . We therefore tested miR-545 expression in glioma tissues and cells. As demonstrated, miR-545 levels are significantly downregulated in glioma tissues (Fig. 1c), as well as in the established and primary human glioma cells (Fig. 1d). In contrast, miR-545 expression is relatively high in normal brain tissues, primary human neuronal cultures and human astrocytes (Fig. 1c, d). Another important miRNA target of circPRKCI, miR-589 16 , was not detected in our system. These results show that circPRKCI is C. D.

Neurons
Astrocytes A172 Pri1 Pri2 Pri3 Neurons Fig. 1 circPRKCI is upregulated in human glioma tissues and cells. Total RNA was extracted from the described human tissues and cells, expression of circPRKCI (a, b) and miR-545 (c, d) was tested by qPCR assays, results were normalized to U6 RNA. "Pat" stands for glioma patient number. Error bars stand for mean ± standard deviation (SD, n = 5). *P < 0.05 vs. "N" tissues (a, c) or primary neuronal culture ("Neurons") (b, d) upregulated in human glioma tissues and cells, correlating with miR-545 downregulation.
circPRKCI shRNA inhibits A172 glioma cell growth, survival, proliferation, and migration In order to study the potential function of circPRKCI in glioma cells, a shRNA strategy was applied. GV248 lentiviral shRNA, targeting non-overlapping sequence ("Seq-1/Seq-2") against circPRKCI ("sh-circPRKCI"), was added to A172 glioma cells. Following puromycin selection stable cells were established. Testing circPRKCI expression in the stable cells, by qPCR assays, confirmed that circPRKCI levels decreased over 90% (vs. the parental control cells) (Fig. 2a). By counting cell number, we show that circPRKCI shRNA-expressing A172 cells grew significantly slower than the control cells (Fig. 2b). Cell viability, or the CCK-8 OD, was decreased as well by circPRKCI shRNA (Fig. 2c). Further experimental results demonstrated that circPRKCI shRNA decreased the number of A172 cell colonies (Fig. 2d) and EdU incorporation (Fig. 2e), indicating its anti-proliferative activity. Expression of proliferation marker proteins, including cyclin D1 and proliferating cell nuclear antigen (PCNA), was significantly downregulated as well in circPRKCIsilenced A172 glioma cells (Fig. 2f).

circPRKCI silencing inhibits subcutaneous A172 glioma growth in SCID mice
To study the potential function of circPRKCI in vivo, stable A172 glioma cells, with circPRKCI shRNA ("Seq-1/ Seq-2") or scramble non-sense control shRNA ("sh-c"), were s.c. injected to the flanks of SCID mice. Within 3 weeks the subcutaneous A172 tumors were established, with tumor volumes close to 100 mm 3 (labeled as Day-0/ "D0"). Recording weekly tumor volumes, in Fig. 5a, demonstrated that A172 xenografts with circPRKCI shRNA grew significantly slower than the control tumors  Fig. 3 circPRKCI silencing inhibits primary human glioma cell progression. The primary human glioma cells, derived from three different patients ("Pri-1/−2/−3"), as well as primary human neuronal cultures ("Neurons") and primary human astrocytes ("Astrocytes"), were transduced with GV248 lentiviral circPRKCI shRNA ("Seq-1") or scramble non-sense control shRNA ("sh-c") for 24 h, stable cells were established via selection with puromycin. circPRKCI expression was tested by qPCR (a); Cell viability (CCK-8 OD, b, f), proliferation (EdU ratio, c), migration ("Transwell" assay, d), and apoptosis (Annexin V ratio, e, g) were tested. A172 cells were transduced with lentiviral circPRKCI expression construct ("LV-circPRKCI") or empty vector ("LV-C") for 24 h, stable cells were established via puromycin selection; circPRKCI expression (h), cell viability (i), and proliferation (j) were tested similarly. Error bars stand for mean ± standard deviation (SD, n = 5). *P < 0.05 vs. "sh-c"/"LV-C" cells. Experiments in this figure were repeated four times, and similar results were obtained (with "sh-c"). When calculating the estimated daily tumor growth, using the formula [tumor volume at Day-35 (D35) subtracting tumor volume at Day-0 (D0)] ÷ 35, we showed again that subcutaneous A172 tumor growth was significantly inhibited by circPRKCI shRNA (Fig. 5b). At D35, tumors of all three groups were isolated and weighted. Results confirmed that circPRKCI shRNAexpressing A172 tumors weighted significantly lower than the control tumors (Fig. 5c). The mice body weights were not significantly different between the three groups (Fig. 5d), neither did we notice any signs of apparent toxicities in the experimental mice.
circPRKCI shRNA inhibits orthotopic A172 glioma growth in mice To further study a role of circPRKCI in glioma cell progression, orthotopic glioma xenografts were established. Exact same number of A172 cells (1 × 10 6 cells of each mouse mouse) with circPRKCI shRNA ("Seq-1/Seq-2") or scramble non-sense control shRNA ("sh-c"), were intracranially injected to the brain of SCID mice 19 . At day-21, with first mouse in the control tumor group showed the typical neurologic and sick symptoms, all groups were sacrificed, with tumors isolated. Results of tumor volumes (Fig. 6a) and the tumor weights (Fig. 6b) demonstrated that orthotopic A172 gliomas with cir-cPRKCI shRNA grew significantly slower than the control tumors (Fig. 6a, b). Once again, the mice body weights were not significantly different between the three groups (Fig. 6c). Biochemical analyses of tumor tissue lysates confirmed that circPRKCI levels were significantly downregulated in the orthotopic A172 xenografts with circPRKCI shRNA (Fig. 6d), whereas miR545 levels significantly increased (Fig. 6e). miR-545 targets, E2F7 and RIG-1, were downregulated in circPRKCI-silenced orthotopic tumor tissues (Fig. 6f). Therefore, circPRKCI shRNA inhibited orthotopic A172 glioma growth in mice.

Discussion
circRNAs are a large and conserved family of noncoding RNAs, generated from a non-canonical back splicing process, from a covalent bond between 5′ and 3′ ends of a single-stranded RNA 14,33 . Dysregulation of cir-cRNAs has been detected in glioma cells, essential for cancer development and progression 12 . Our results suggest that circPRKCI is an oncogenic circRNA in glioma. Its expression is upregulated in human glioma tissues and in established/primary human glioma cells, but low in normal brain tissues and neurons/astrocytes. In A172 cells and primary human glioma cells, circPRKCI silencing, by targeted shRNA, potently inhibited cell growth, survival, proliferation, and migration, whiling inducing apoptosis activation. Conversely, exogenous overexpression of circPRKCI, by a lentiviral construct, promoted A172 cell progression. Furthermore, in vivo growth Fig. 5 circPRKCI silencing inhibits subcutaneous A172 glioma growth in SCID mice. Weekly tumor growth curve of subcutaneous A172 gliomas with circPRKCI shRNA ("Seq-1/Seq-2") or scramble non-sense control shRNA ("sh-c") was shown (a); Estimated daily tumor growth was calculated using the described formula (b); At D35, tumors of all three groups were isolated and weighted (c); Mice body weights were recorded (d). At D7 and D14, one tumor of each group was isolated (total six tumors). Each tumor was randomly cut into five small pieces (n = 5), individually dissolved into the lysis buffer. Expression of circPRKCI (e), miR-545 (f), and listed proteins (g) in each piece was tested separately. Results were integrated and statistics analyses were performed (e, f). Representative blotting data were shown (g). Error bars stand for mean ± standard deviation (SD). *P < 0.05 vs. "sh-c" tumors of subcutaneous and orthotopic A172 gliomas was significantly inhibited by circPRKCI silencing. These results suggest that circPRKCI could be an important and novel therapeutic target of glioma.
Recent studies have shown that miR-545 is a tumorsuppressive miR in human cancers 16,28,29 . Du et al., have shown that miR-545 can inhibit human lung cancer cells 29 . In human pancreatic ductal adenocarcinoma cells, miR-545 inhibits RIG-1 expression and cancer cell growth 28 . In lung cancer cells circPRKCI promotes cancer cell progression by sponging miR-545 16 . In this study, we show that miR-545 inhibition or knockout promoted A172 cell progression. Conversely, forced overexpression miR-545 inhibited A172 cell survival and proliferation. Therefore, miR-545 plays a tumor-suppressive role in human glioma cells.
Two primary miR-545 targets are RIG-1 and E2F7 in cancer cells 16,26,28,34 . RIG-I is an intracellular viral RNA sensor, whose activation could initiate host innate immune response to increase type I IFN production 35 . Recent studies have proposed a role of RIG-1 in cancer progression. For instance, Song et al., demonstrated that high RIG-I protein level in pancreatic ductal adenocarcinoma tissues is correlated with shorter survival 28 . Two cancer studies have discovered that miR-545 inhibits human cancer progression by targeting RIG-I 28,34 . E2F7 is a relatively novel transcription factor, regulating cell cycle by inhibiting expression of G1-S genes 36  D. E. F. Fig. 6 circPRKCI shRNA inhibits orthotopic A172 glioma growth in mice. Same number of A172 glioma cells, expressing circPRKCI shRNA ("Seq-1/Seq-2") or scramble non-sense control shRNA ("sh-c"), were intracranially injected into the brain of SCID mice (5-6 week old, eight mice per group), after 21 days the animals were sacrificed and tumor volumes (a), tumor weights (b), and mouse body weights (c) recorded. Tumor tissue lysates were analyzed by qPCR and western blotting assays to test listed genes (d-f). E2F7 and RIG-1 protein expression was quantified (f). Error bars stand for mean ± standard deviation (SD, n = 5). *P < 0.05 vs. "sh-c" tumors form a heterodimer with E2F1 and recruits the corepressor C-terminal-binding protein (CtBP) 16 . Recent studies have implied that downregulation of E2F7 could induce cancer cell apoptosis 38,39 . Importantly, E2F7 could be an independent prognostic factor of gliomas, whose overexpression predicts poor prognosis in glioma patients 40 . Thus, E2F7 could be a novel therapeutic target of human glioma 40 . E2F7 is one key mRNA target of miR-545 16 . Here, in A172 and primary human glioma cells ectopic miR-545 overexpression significantly downregulated RIG-1 and E2F7, both were upregulated with miR-545 inhibition. circPRKCI could sponge miR-545 and possible other tumor-suppressive miRNAs 16 . The results of the current study indicate that miR-545 is the primary target of cir-cPRKCI in glioma cells. RIP assay results show that cir-cPRKCI and miR-545 were both efficiently pulled down by anti-Ago2 antibody in A172 glioma cells. miR-545 levels were significantly increased in circPRKCI-silenced A172 cells, with its targets, E2F7 and RIG-1, downregulated. Importantly, exogenous overexpression of miR-545 by a lentiviral construct potently inhibited A172 cell progression, mimicking circPRKCI shRNA-induced activity. Conversely, miR-545 inhibition, via LV-antag-omiR-545, abolished circPRKCI silencing-induced anti-A172 cell activity. Significantly, miR-545 inhibition or knockout (by CRISPRC/Cas9 method) promoted A172 cell progression. Remarkably, neither circPRKCI shRNA nor circPRKCI overexpression was effective in the miR-545-KO A172 cells. In the circPRKCI-silenced subcutaneous and orthotopic A172 xenograft tumor tissues, miR-545 levels were significantly upregulated, correlating with downregulation of its targets, RIG-1 and E2F7. Finally, we show that in human glioma tissues and cells, circPRKCI upregulation correlates with miR-545 downregulation. These results indicate that circPRKCI promotes glioma cell progression possibly by sponging miR-545. miR-545 should be the direct target of circPRKCI in glioma cells.

Conclusion
circPRKCI promotes human glioma cell progression possibly by inhibiting miR-545. Targeting circPRKCI-miR-545 cascade could be a novel strategy to inhibit human glioma.