miR-429 Suppresses Proliferation and Migration in Glioblastoma Cells and Induces Cell-cycle Arrest via Modulating Several Target Genes of ERBB Signaling Pathway

Glioblastoma is aggressive and lethal brain cancer, which is incurable by cancer standard treatments. miRNAs have great potential to be used for gene therapy due to their ability to modulate several target genes simultaneously. We found miR-429 is downregulated in glioblastoma and has several predicted target genes from the ERBB signaling pathway using bioinformatics tools. ERBB is the most overactivated genetic pathway in glioblastoma patients, which is responsible for augmented cell proliferation and migration in glioblastoma multiforme (GBM). Here we overexpressed miR-429 using lentiviral vectors in GBM U-251 cells and observed that the expression level of several oncogenes of the ERBB pathway, EGFR, PIK3CA, PIK3CB, KRAS, and MYC signicantly decreased; as shown by real-time PCR and western blotting. Using the luciferase assay, we showed that miR-429 directly targets MYC, BCL2, and EGFR. In comparison to scrambled control, miR-429 had a signicant inhibitory effect on cell proliferation and migration as deduced from MTT and scratch wound assays and induced cell-cycle arrest in ow cytometry. Altogether miR-429 seems to be an ecient suppressor of the ERBB genetic signaling pathway and a potential therapeutic for glioblastoma.


Introduction
Glioblastoma multiforme (GBM) is the most frequent type of malignant brain and other CNS tumor in adults (14.6% of all tumors and 48.3% of malignant tumors) 1 . About 3.2 per 100,000 people are diagnosed with glioblastoma multiforme annually with a ve-years survival rate of 6.8% post-prognosis in the case of receiving therapy 1 . Fast growth and high mobility of glial cells 2 , genetic heterogeneity of glioblastoma tumors 3 , presence of stem-like cancer cells 4 , and blood-brain barrier (BBB) that limits the immune system to function in the brain 5 , made glioblastoma highly lethal and incurable with conventional treatments of cancer such as chemotherapy and radiotherapy 6 . Poor prognosis and high recurrence rate of glioblastoma highlight the urgent need for novel therapeutic strategies.
Gene-therapy is promising for cancer treatment [7][8][9] . Various gene-therapies are in clinical trial phases for glioblastoma, which aims to induce the expression of therapeutic genes, such as tumor suppressor, suicide, and immunostimulatory genes, or suppress the expression of oncogenes 10 . miRNAs are natural 22-24 nucleotides long oligonucleotides, which play a role in genetic network regulation at the translation level by binding to complementary regions of 3'-UTRs of mRNAs and blocking their translation 11 . miRNAs are great candidates in gene therapy owing to their ability to suppress the expression of genes of interest [12][13][14] . miRNAs expression level changes during different physiological and pathological conditions of the body and directly correlates with changes in their target genes expression pro les [15][16][17][18][19] . We can also change the miRNAs' target genes' level by exogenous expression of miRNAs for therapeutic means [20][21][22][23] .
To date, about 140 genetic mutations have had a role in glioblastoma multiforme progression 24 . GBM patients usually have more than one mutation and sometimes hypermutations (about 60 mutations per tumor) 25 . EGFR (epidermal growth factor receptor), a transmembrane glycoprotein, functioning as a receptor tyrosine kinase in the ERBB signaling pathway, is ampli ed in up to 60% of GBM patients showing to have a role in cell proliferation, growth, and survival 26 . The activation of EGFR by binding to its ligand triggers its downstream pathways such as phosphatidylinositol-3-kinase (PI3K)/ Protein Kinase B (AKT) and the mammalian target of rapamycin (mTOR) or RAS/RAF/MAPK (mitogen-activated protein kinases) 26 . Both pathways are also highly activated in GBM and subject to other activating mutations in GBM patients, including PDGFRA (10%), FGFR (3.2%), PI3K (25%), PTEN (41%), NF1 (10%), KRAS (1%) and BRAF (2%) 27 . ERBB signaling activation leads to the induction of tumor progression, invasion, angiogenesis, and chemotherapy resistance 28 . The use of miRNAs to target ERBB pathway oncogenes decreased proliferation of GBM cells in vitro 29,30 .
Here we overexpressed miR-429 in the GBM U-251 cell line and studied its effect on direct and indirect regulation of several ERBB signaling pathway oncogenes, cell proliferation, migration, and apoptosis rate.

In silico miRNA/target selection
We used the miRWalk 2.0 online tool 31,32 , the Gene-miRNA-pathway tab (http://zmf.umm.uniheidelberg.de/apps/zmf/mirwalk2/path-self.html) to predict miRNAs which suppress the ERBB signaling pathway. We chose miR-429 as a candidate with several oncogenes from the ERBB pathway predicted to be targeted by that; so we analyzed a miRNA microarray dataset (GSE90603) from the GEO database (Gene Expression Omnibus from NCBI website) 33

MiRNA cloning
We obtained the sequence of miR-429 stem-loop from the miRBase website 42,43 (http://www.mirbase.org/). Using a nucleic acid editing tool, from the Genome Data viewer (NCBI website) (https://www.ncbi.nlm.nih.gov/genome/gdv/), 200 nucleotides were added to each side of the stem-loop to assure its right folding and make its cloning easier. We designed speci c primers and added EcoRI and BamHI restriction enzymes sites to them, after checking that they won't cut the sequence desired for cloning. Forming the right miR-429 stem-loop was seen in the RNAfold WebServer online tool (http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi). We performed the polymerase chain reaction (PCR) and then digested the PCR product by restriction enzymes (Thermo Fisher Scienti c, Waltham, MA). They were ligated by T4 DNA Ligase (Thermo Fisher Scienti c) to pCDH-GFP-Puro

Viral packaging and transduction
We cotransfected the seeded HEK293T cells by lentiviral vector pCDH-GFP-Peuro-miR-429/ scrambled, psPAX2 packaging vector, and pMD2G-VSVG vector with polyethyleneimine (Sigma). For four days, cell supernatants containing secreted recombinant viruses were harvested every 24 hours and stored at 4°C adding new media to cells. On the fourth day, cell debris was eliminated from viral supernatants by centrifuge at 2000 X G /4°C for 10 minutes, followed by ltering with 0.2 µm syringe lters. Viral supernatants were aliquoted and stored at -80°C before use. U-251 glioblastoma cells were seeded in appropriate sterile dishes (according to the following cellular assay) at 24 hours before transduction. We used 10 µg/ml polybrene (Sigma) to enhance the transduction of virus-containing supernatants. Fluorescent microscopy assured the e ciency of transduction at 48 hours through GFP detection. We performed the cellular assays in a minimum of 90% e ciency of transduction. Otherwise, Puromycin 1 µg/ml (Sigma) was added for 48 hours to enrich and select transduced cells.

Gene expression by real-time PCR
Total RNA with miRNAs were extracted from U-251 cells using TRIzol (Invitrogen) on 72 hours after transduction with miR-429/scrambled viruses and then measured by spectrophotometer (Eppendorf). 5µg of each RNA went through complementary DNA (cDNA) synthesis reactions with random hexamer primer (for total mRNA) or speci c stem-loop primers (for miRNAs), by using M-MuLV Reverse Transcriptase enzyme (Thermo Fisher Scienti c), based on manufacturer's instruction. Design of RT-stem loop primers, as well as miRNA forward and reverse primers, was performed based on a previously published method 44 for miR-429 and SNORD47. Speci c primers were designed for evaluation of target genes expression via Real-time PCR, using the Primer-Blast online tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/). We performed real-time PCR with SYBR Green master mix 2X (Ampliqon, Odense M, Denmark) based on the manufacturer's instruction via ABI 7500 (Applied Biosystems, USA) machine. The quantitative PCR program was 5 minutes of 95°C followed by 40 cycles of 95°C for 15 seconds and 62°C for 1 minute. 2 −ΔΔCt method was used to calculate expression fold changes of miR-429 and some of its predicted target genes, normalized to β2M and SNORD47 as internal controls for mRNAs and miRNAs respectively. PCR reactions were duplicate; we experimented with three biological repeats. Used Primers are in Tables 1 and   2. and anti-β-Actin (1:200; Abcam) mouse monoclonal antibodies. We washed the membrane with PBS-Tween buffer and exposed it to the secondary antibody solution (1:1000, Abcam) conjugated with horseradish peroxidase. We captured a photo after the addition of ECL Western blot analysis substrate (Thermo Fisher Scienti c) to the membrane in the dark, and the density of bands was investigated using GelAnalyzer software 2010a and normalized to ACTB as an internal control.

Dual-luciferase reporter assay
We designed speci c primers containing restriction enzyme sites at their 5' to amplify mRNA 3'untranslated regions (  incubation in the dark. Then cells were washed another washing step by 1x binding buffer, stained by adding ve µl 7AAD, and right after were subject to ow cytometry. Obtained data were analyzed using FlowJo 7.6.1 software. We experimented with three biological repeats. 2.11. Cell-cycle analysis assay U-251 cells were seeded and transduced in a 24-well plate (4 x 10 4 cells/well) with miR-429/ scrambled viruses (2 replicates for each). At 72 hours after transduction, we harvested the cells by trypsinization. Then washed them with PBS, gently add them to another microtube containing cold 70% Ethanol on vortex. Fixed cells were kept in the fridge for at least 4 hours before cell-cycle assay. Then we centrifuged the cells, removed Ethanol, and washed the cell pellet once with PBS. We stained cells by adding 200 µl of Propidium Iodide (PI; 50 µg/mL) (Sigma), RNase (1.0 mg/mL) (Thermo Fisher Scienti c), and Tryton X-100 (Sigma) followed by 40 minutes' incubation at 37°C in the dark. Then we performed Flow cytometry (BD Biosciences) on cells and used FlowJo 7.6.1 software for the cell-cycle analysis of the data. We experimented with three biological repeats.

Statistical analysis
Biological repeats of the experiments were statistically analyzed using GraphPad Prism 7.04 (San Diego, CA) software. The signi cancy of the data between miR-429-treated and scrambled-treated groups was investigated by applying Student's t-test. We presented the data as mean ± standard deviation. P < 0.05 was considered a statistically signi cant change.

Results
3.1. miR-429 is downregulated while its predicted target genes from the ERBB pathway are upregulated in glioblastoma tissue samples in silico First, we investigated the expression changes of miR-429 in glioblastoma patients' tumors versus normal tissues based on the GSE90603 miRNA microarray dataset obtained from the GEO database. miR-429 expression data of 16 tumor tissue and four healthy tissue samples from GBM patients showed signi cant downregulation of miR-429 in glioblastoma ( Fig. 1.A). The differential expression of miR-429 in glioblastoma tumors is 0.67 ± 0.04 versus 0.88 ± 0.11 in normal tissues (*P < 0.05).
Next, we predicted targets of miR-429 in the ERBB pathway using the miRWalk, and TargetScan on-line tools. Predicted targets of miR-429 in TargetScan/ miRWalk and their commonality with ERBB signaling pathway genes (from KEGG database) are shown in the Venn diagram using Bioinformatics & Evolutionary Genomics website ( Fig. 1.B, Table 3). Based on TargetScan, 15 target genes, and miRWalk, six target genes of miR-429 are members of the ERBB signaling pathway; Three target genes are in both datasets. The miR-429 predicted target genes in the ERBB pathway are upregulated based on the Pan-Cancer analysis of the whole genome-brain on the Expression Atlas database (Fig. 1.C). According to the TCGA-GBM project in the TCGA website ( Fig. 1.D), these genes are subject to alterations (mutations or copy number variations) in glioblastoma cases. These ndings suggest that miR-429 predicted target genes have a role in GBM progression.

miR-429 modulates several ERBB target genes
To verify the in uence of miR-429 on its predicted target genes from the ERBB pathway, we overexpressed miR-429/ scrambled in the U-251 glioblastoma cell line using lentiviral transduction. Overexpression of miR-429 was con rmed by detecting GFP by uorescent microscopy in 48 hours (Fig. 2.A) and by real-time PCR in comparison to scrambled in 72 hours after transduction. miR-429 was 28.81 ± 3.13 fold overexpressed (**P < 0.01) after transduction with miR-429 Lentivirus compared to scrambled control (Fig. 2.B).

miR-429 directly targets MYC, BCL2, and EGFR
To verify the direct target genes of miR-429, we cloned 3'-UTRs of three target genes, MYC, BCL2, and EGFR in the psiCHECK2.0 vector, downstream of the Renilla luciferase coding sequence. Then, we cotransfected miR-429/ scrambled and 3'-UTR vectors to HEK293 cells. At 48 hours, we performed the dualluciferase reporter assay measuring the luciferase activity. The relative luciferase activity was decreased signi cantly (*P < 0.05) in all three cases (Fig. 3.A, 3.B). This result approves direct targeting of these three genes, i.e., MYC, BCL2, and EGFR by miR-429.

miR-429 overexpression leads to proliferation and migration suppression in U-251 cells
To investigate the biological function of miR-429, we performed MTT proliferation assay and scratch assay after overexpression of miR-429 in U-251 glioblastoma cells. MTT assay at 72 hours after transduction showed a signi cant decrease in viable cells, suggesting that cells had a diminished proliferation rate. The viability of miR-429 transduced cells was 85.87 % ± 6.32 (**P < 0.01) related to the control group (Fig. 4.A).
In another plate, we made a scratch at 48 hours after transduction, and the closure rate of cells was measured 0, 24, 48, and 72 hours after scratch, as an indicator of migration potential. After 72 hours' closure rate was signi cantly decreased (*P < 0.05) in miR-429 transduced cells in comparison to control ( Fig. 4.B, 4.C). So, our results from biological assays suggest miR-429 potential role to inhibit glioblastoma cell proliferation and invasion in vitro.

miR-429 overexpression does not cause apoptosis in U-251 cells, but cell-cycle arrest
To investigate the miR-429 effect on glioblastoma cells apoptosis, we performed Annexin-PE/7AAD assay 72 hours after transduction. Despite the decrease in cell proliferation and migration, miR-429 did not seem to cause any signi cant effect on apoptosis (early and late) during 72 hours after transduction, as measured by ow cytometry (Fig. 5.A, 5.B, 5.C).

Discussion
In recent years, miRNAs have attracted scientists' attention because of their capacity to regulate target genes involved in cancer development and progression. Therefore, their implementation as a tool to suppress cancer cell proliferation and invasion either by induction of apoptosis or cell cycle arrest is raised 13,45,46 . The promising outcome of clinical trials in the application of miRNAs in cancer treatment like anti-miR-122, Miravirsen, has shown the great potentials of miRNAs as therapeutic tools 12 . Herein we showed that the Lentiviral delivery of miRNAs to glioblastoma cells is very e cient. Stable transduction using Lentiviral delivery provides permanent miRNA gene expression via antibiotic selection methods.
U-251 glioblastoma cell line used in this study is a typical model of glioblastoma tumor cells that carries a wide range of genetic mutations. U-251 is mutant for TP53 (encoding P53, a molecular inducer of apoptosis), CDKN2A (encoding P16 and P14, molecular inducers of cell cycle arrest), PTEN (encoding inhibitor molecule of PI3K), NF1 (encoding inhibitor molecule of RAS), and EGFR (encoding epidermal growth factor receptor) 47,48 . EGFR, RAS, and PI3K are members of the ERBB signaling pathway, which is amongst the most overactivated genetic mechanisms in glioblastoma patients, leading to cell proliferation and invasion 49 . So a miRNA that targets and suppresses the ERBB signaling pathway was predicted to be a potential tumor suppressor miRNA.
Our bioinformatical analysis predicted that miR-429 could target several members of the ERBB pathway.
These molecular targets were upregulated in glioblastoma tissues comparing to the normal brain in silico and were also subject to alterations (mutations or copy number variations) in glioblastoma patients. Besides, miR-429 expression was reduced in glioblastoma patients samples. These ndings suggested that overexpression of miR-429 in glioblastoma cell lines might modulate some of these oncogenes from the ERBB pathway directly or indirectly, so suppress glioblastoma tumor cells proliferation and migration. Our results for MTT proliferation assay and scratch wound assay of U-251 GBM cells con rmed the tumor suppressor effect of miR-429 in glioblastoma. Although miR-429 suppressed proliferation and migration in U-251 cells, it did not induce apoptosis. So we speculate it may induce differentiation in GBM cells with suppressed proliferation, which is along with our results on the cell-cycle arrest of GBM cells by miR-429 come from ow cytometry analysis of cell-cycle. Our previously published data con rm that the neuronal differentiation genetic markers increase by miR-429 overexpression in GBM cells 59  Besides, our approach to using miRNAs to modulate a genetic pathway instead of a unique target gene, especially in genetically heterogeneous cancers like glioblastoma, seems to make the bioinformatical prediction more realistic and e cient in vitro or in vivo. Our ndings, especially the miR-429 regulatory effect on the ERBB signaling pathway in glioblastoma, suggests its potential role as a therapeutic agent not only in glioblastoma but also in any other cancer or diseases, in which ERBB signaling pathway over activation is the main factor.

Declarations
The authors declare that there is no con ict of interest.