Disturbing miR-182 and -381 Inhibits BRD7 Transcription and Glioma Growth by Directly Targeting LRRC4

Inactivated LRRC4 has been clinically detected in gliomas, and promoter hypermethylation has been implicated as the mechanism of inactivation in some of those tumors. Our previous researches indicated that LRRC4 is a target gene of miR-381, the interaction of miR-381 and LRRC4 is involved in glioma growth. In this study, we demonstrate that LRRC4 is a target gene of the other microRNA, miR-182. We found that the high expression of miR-182 and miR-381 in gliomas are involved in pathological malignant progression. The silencing of miR-182 and miR-381 inhibited the proliferation in vitro and growth of glioma cell with in vivo magnetic resonance imaging by intracranial transplanted tumor model in rats. We also demonstrated that BRD7, a transcriptional cofactor for p53, is highly expressed and negatively correlated with LRRC4 expression in gliomas. Disturbing miR-182 and miR-381 affected transcriptional regulation of the BRD7 gene. This finding was verified by ectopic overexpression of LRRC4 or restoration of endogenous LRRC4 expression by treatment with the DNA demethylating agent 5-Aza-dC. Taken together, miR-182 and miR-381 may be a useful therapeutic target for treatment of glioma.


Introduction
Gliomas are the most common type of primary brain malignancies and are characterized by well-defined genetic alterations and frequent chromosomal aberrations that are associated with tumorigenesis [1].The chromosome region 7q31-32 is thought to contain multiple tumor-suppressor genes involved in the pathogenesis of head and neck tumors, one of which is the leucine-rich repeat C4 (LRRC4) gene [2]. Our previous studies indicated that LRRC4 gene expression was highly specific to brain tissue [3], and that the gene product behaved as a tumor suppressor in the pathogenesis of malignant gliomas [4]. Inactivated LRRC4 has been clinically detected in gliomas, and promoter hypermethylation has been implicated as the mechanism of inactivation in some of those tumors [5]. When LRRC4 expression was induced in the human U251 glioma cell line, the Ras/p-c-Raf/ERK/MAPK and PI-3K/AKT signaling pathways were found to be down-regulated, which in turn inhibited cell proliferation and invasion [6,7][.Functional studies of mouse LRRC4 (mLRRC4) determined that the gene was crucial for fetal brain development, as it mediated cell cycle progression and neuron and glia cell differentiation [2,8]. Therefore, LRRC4 is considered an important regulator of cell maintenance, growth, and differentiation in the nervous system. However, the molecular mechanism by which LRRC4 regulates glioma tumorigenesis has not been fully elucidated.
An increasing number of studies have demonstrated that tumour suppressive miRNAs are frequently downregulated in tumours, whereas oncogenic miRNAs are frequently upregulated [9]. We previously reported that miR-200b is greatly downregulated in gastric cells and tissues and inhibits gastric cancer proliferation and invasion [9,10]. In contrast, the oncogenic miR-183/182/96 cluster of miRNAs is upregulated in a variety of tumours [9,11,12], and we previously reported that miR-183/182/ 96 cluster regulates oxidative apoptosis and sensitises gliomal cells to chemotherapy [13]. miRNA-mediated gene regulation has been demonstrated to contribute to numerous life-sustaining biological processes, including cell proliferation, differentiation, development, and metabolism, and to pathogenic processes, such as cancer formation [14][15][16].Computational algorithms have been developed to identify potential gene targets of miRNAs, either by using the miRNA to find genes that harbor the predicted target sequence or by using a known gene to search for a miRNA that would match its nucleotide sequence. When the glioma-related LRRC4 gene was queried by TargetScan and PicTar software, it was identified as a target gene of miR-182 and miR-381. We have confirmed that LRRC4 is a target gene of miR-381, at the same time, overexpression of LRRC4 also downregulated the expression of miR-381 in glioma cells, the interaction of miR-381 and LRRC4 is involved in glioma growth [17]; when the miRNAs were used for query, it was found that miR-381 targets both LRRC4 and BRD7, which is cloned by our laboratory [18].
In our laboratory, BRD7 has been reported as absent or expressed at unusually low levels in nasopharyngeal carcinoma (NPC) biopsies, and overexpression of BRD7 inhibits NPC cell growth and arrest cells in the G0/G1 phase of the cell cycle [18]. Within the BRD7 promoter, a Sp1/Myc-Max overlapping site has been defined as a key regulatory element; Sp1 binding leads to slightly positive regulation of promoter activity, while c-Myc binding negatively regulates activity [19]. Moreover, in other laboratories, BRD7 was found to act as a transcriptional cofactor for p53 [20], and BRD7 is a candidate tumour suppressor gene required for p53 function [21]. BRD7 interacts with PRMT5 and PRC2, and is involved in transcriptional repression of their target genes [22]. But, the expression and function of BRD7 in gliomas have yet to be reported.
On the basis of our previous research that LRRC4 is a target gene of miR-381, we confirmed LRRC4 is also the target gene of miR-182. BRD7 isn't the target gene of miR-381, but miR-381 did downregulated the expression of BRD7. We further found that the expression of miR-182 and miR-381 or BRD7 and LRRC4 were negatively correlated with the pathological progression of gliomas. Both ectopic LRRC4 overexpression, or the restoration of LRRC4 endogenous expression inhibited the expression of BRD7, locked-nucleic-acid (LNA)-mediated miR-182 and -381 silencing specifically targeted the LRRC4 and led to perturbations in the Ras/Raf/ERK/MAPK and PI-3K/AKT signaling pathways and to down-regulation of BRD7, which inhibited glioma growth. The study not only demonstrated the mechanism of tumor suppressor gene LRRC4 involved in the glioma progression, but also provided the therapy targets for glioma.

LRRC4 is a bona fide target of miR-182
The miRNA target prediction programs TargetScan and PicTar identified miR-182 interaction sites in the 39-UTR of LRRC4 ( Figure 1A). We confirmed that LRRC4 is a common bona fide target of miR-182 ( Figure 1B) by performing luciferase reporter assays. LRRC4 is specifically expressed in normal brain tissues and is down-regulated or deleted in primary brain tumor biopsies (up to 87.5% in gliomas) and glioma cell lines [4]. Because cell lines expressing high levels of LRRC4 are difficult to obtain, we chose to use U251 cells stably transfected with LRRC4 to generate a model of LRRC4 highly expressing cells (U251/L cells) [4]. Transfection of the miR-182 mimics into U251/L cells resulted in a marked reduction of LRRC4 at both the protein ( Figure 1C) and mRNA levels, as compared with the control (scrambled) transfection ( Figure 1D). Although BRD7 was predicted as a putative target of miR-381, the miRNA did not combine with the 39-UTR of BRD7 and the gene sequence was not a bona fide target. However, overexpression of miR-381 mimics did downregulated the expression of BRD7 ( Figure S1). miR-182, miR-381, BRD7 are inversely correlated with LRRC4 expression in gliomas To evaluate the relevance of the endogenous expressions of miR-182, miR-381, BRD7, and LRRC4, we assessed their expressions in human glioma tissues, as well as in normal brain tissues. ISH analysis of miR-182 and miR-381 and IHC analysis of BRD7 and LRRC4 showed that all 47 primary gliomas had elevated levels of miR-182, miR-381, and BRD7 (except for one case), and decreased levels of LRRC4 (except for two cases), as compared to levels detected in the 21 normal brain tissues. We found that the expressions of miR-182, miR-381 or BRD7 proteins were inversely correlated with expression of LRRC4 in glioma tissues and normal brain tissues ( Figure 2A). More importantly, we also found that miR-182, miR-381 and BRD7 were inversely correlated with LRRC4 in astrocytomas of various pathological grades ( Figure 2B, C), and the extent of correlation increased from WHO grade I to IV. In contrast, the expression of LRRC4 in astrocytomas gradually decreased with increase in WHO grade, from I to III, and was completely absent in grade IV astrocytomas ( Figure 2B, C). In addition, there was a positive correlation found between expressions of miR-182 or miR-381 and BRD7 in all glioma tissues, and normal brain tissues. As shown in Figure S2, qRT-PCR was used to furhter verify the expression levels of miR-182 and miR-381 in 19 normal brain tissue samples and 67 primary gliomas. The results indicated that expression change of miR-182 and miR-381 in normal brain tissue and different grade gliomas (I: 10 cases; II: 22 cases; III: 23 cases; IV: 12 cases) was consistent to that detected by ISH (Figure 2 A-C).

miR-182 and miR-381 silencing inhibited glioma tumorigenicity and induced differentiation
To directly test the functional roles of miR-182 and miR-381 in cell proliferation, ectopic miR-182 and miR-381 mimics or LNAanti-miR-182 and -381 oligonucleotides were transfected into multiple glioma cell lines. There were obvious increases in the proliferation of U251, U87, and P19 cells by 24 h after transfection with the miR-182 and -381 mimics. and an increase in the proliferation of SF126 and SF767 cells by 48 h, as compared with the control (mock-transfected) cells and cells transfected with a scrambled sequence (t-test, p,0.01; Figure 3A). In contrast, LNA-anti-miR-381 and miR-381 mimics transfection significantly decreased the proliferation of all of these cell lines within 48 h ( Figure 3A). Similar patterns were observed for ectopic miR-182 expression or LNA-mediated miR-182 silencing (data not shown). A cell-cycle analysis showed that transfection of LNAanti-miR-182 and -381 into U251 cells led to a decrease in the proportion of cells in the S and G2-M phases and a corresponding increase in cells in the G1 phase. The LNA-anti-miR-182 and -381 transfections also increased Rb expression and decreased E2F3 expression ( Figure 3B). Ectopic LRRC4 expression [2][3][4], 5-Aza-dC treatment [5], and BRD7-siRNA transfection each inhibited cell proliferation and induced cell cycle arrest at G0/ G1 in glioma cells ( Figure 3C). However, compared with treatment with either LNA-anti-miR-182, LNA-anti-miR-381 or 5-Aza-dC alone, the combination treatment of LNA-anti-miR-182/5AZa, LNA-anti-miR-381/5AZa, or LNA-anti-miR-182/ LNA-anti-miR-381 did not induce any obvious differences in the growth inhibition of U251 cells ( Figure S3).
Encouraged by these in vitro results, we investigated the feasibility of using the LNA-anti-miR-182 and -381 oligonucleotides as an in vivo anti-tumor agent. U251 cells were subcutaneously injected into nude mice to yield tumors that were then treated by direct intratumoral injection as soon as they became palpable. The growth curves of treated vs. control-treated tumors are compared in Supplementary Figure S1B. As shown, the growth curves slowly diverged, reaching statistical significance at 21 days after the first antagomir injection (t-test, p,0.01).
Treatment of U251 cells with LNA-anti-miR-182 or -381 led to differentiation into astrocyte-like cells, as demonstrated by the induced expression of glial fibrillary acidic protein (GFAP). In contrast, transfection of the LNA-scrambled control resulted in a low level of GFAP expression. There were no differences in the levels of detected GAPDH or 49, 69-diamidino-2-phenylindole miR-182 and miR-381 Are Targets for Glioma Therapy (DAPI) nuclear counterstaining between the different groups ( Figure 3D).
Next, we investigated whether the LNA-anti-miR-182 and/or -381 was able to surpass the blood-brain barrier and inhibit growth of the intracranial transplanted tumors. C6 glioma cells tumors were intracranially transplanted into the brain parenchyma of Sprague-Dawley rats. The transplanted rats were treated with intraperitoneal injections of 200 mL of 0.9% saline solution containing: 2 mg of scrambled, LNA-anti-miR-182, or LNA-anti-miR-381, or 1 mg of LNA-anti-miR-182+1 mg of LNA-anti-miR-381 over one day. Magnetic resonance imaging (MRI) revealed that administration of miR-182 or miR-381 inhibitors was accompanied by significantly reduced growth of the intracranial transplanted tumors. The inhibitory effect of the combination treatment with miR-182+miR-381 inhibitors was more robust than single treatments ( Figures 4A and S4A). ISH and IHC indicated that LNA-anti-miR-381 and/or LNA-anti-miR-182 treatment increased the expression of LRRC4 and reduced the expression of BRD7 ( Figures 4B and S4B).

miR-182 and miR-381 silencing induced BRD7 downregulation by targeting LRRC4
To further test the effects of miR-182 and miR-381 on endogenous LRRC4 and BRD7 expressions, we used simple systemic delivery of an unconjugated, PBS-formulated LNA-anti-miR to antagonize expression of endogenous miR-182 and -381 in U251 cells ( Figure 5A). Western blotting indicated that LNAmediated silencing of miR-182 and -381 led to recovery of LRRC4 expression and reduced BRD7 expression ( Figure 5B). These results were confirmed by qRT-PCR ( Figure 5C).
Previous studies had indicated that LRRC4 was down-regulated in glioma biopsies and cell lines [2,4], and that methylation of the LRRC4 promoter was one of the reasons for LRRC4 inactivation in some gliomas [5]. To test if the observed down-regulated of BRD7 was LRRC4-dependent, we analyzed the effects of ectopic LRRC4 expression and 5-Aza-dC demethylation treatment on BRD7 expression. Ectopic LRRC4 overexpression in U251 cells ( Figure 5D) and 5-Aza-dC-mediated re-expression of LRRC4 in U251, SF126, and SF767 cells ( Figure 5E) led to suppression of BRD7 expression, suggesting that miR-182 and miR-381 silencing inhibited BRD7 expression by targeting LRRC4.
miR-182 and miR-381 silencing regulated AP2/SP1/E2F6/ c-Myc-mediated BRD7 transcription induced by LRRC4mediated K-Ras/c-Raf/ERK/MAPK and PI-3K/AKT signaling pathways Previous studies have demonstrated that LRRC4, a regulator of miR-182 and miR-381, can inhibit glioma tumorigenicity by modulating receptor tyrosine kinase (RTK) signaling pathways, such as the K-Ras/p-c-Raf/ERK/MAPK and PI-3K/AKT signaling pathways [4,6]. Therefore, we investigated the signal transduction pathways involved in BRD7 expression that was inhibited by miR-182 and -381 silencing. Specifically, we analyzed the expression and/or the activation of some of the proteins involved in the K-Ras/p-c-Raf/ERK/MAPK and PI-3K/AKT signaling pathways in response to miR-182 and miR-381 silencing. The Western blot results showed that transfection of LNA-anti-miR-182 and -381, but not of the LNA-scrambled controls, inhibited the expression of K-Ras, p-c-Raf, pERK, PI-3K, and pAKT, but had no effects on N-Ras, total ERK, and AKT expressions ( Figure 6A).
Previous studies have also identified some critical transcription factor binding sites in the BRD7 promoter, namely those for AP2, c-Myc, E2F6, E2F, and Sp1, and confirmed that BRD7 transcription is positively regulated by SP1 and negatively regulated by c-Myc. We investigated whether miR-182 and miR-381 silencing could induce these transcription factors. First, we examined the effects of different kinase inhibitors (PD98059 for MEK, and LY294002 for PI-3K) on the miR-182-and miR-381induced expression of the transcription factors. Our results indicated that, compared with the LNA-anti-scrambled control, transfection of LNA-anti-miR-182 and -381 increased and  Figure 6B). Treatment with PD98059 or LY294002 abrogated the effects induced by the miR-182 and miR-381 mimics ( Figure 6C). In addition, to determine whether the miR-182 and miR-381 silencing-induced effects on the expressions of AP2, SP1, E2F6, and c-Myc were LRRC4dependent, we performed LRRC4 overexpression studies. We found that ectopic overexpression of LRRC4 in U251 cells ( Figure  S5A), and endogenous expression of LRRC4 induced by 5-Aza-dC in U251, SF126, and SF767 cells (Supplementary Figure S5B), resulted in decreased expression of K-Ras, p-c-Raf, pERK, PI-3K, and pAKT, and reversed the expression of AP2, SP1, E2F6, and c-Myc transcription factors. These results suggested that miR-182 and miR-381 silencing modulated the ERK/MAPK and PI-3K/ AKT signaling pathways, thereby inhibiting the expression of AP2, SP1, and E2F6 and promoting c-Myc expression.
Next, we investigated the effects of miR-182 and miR-381 silencing on the ability of AP2, SP1, E2F6, and c-Myc to bind to the promoter of BRD7. As evidenced by EMSA, transfections of LNA-anti-miR-182 or -381 disrupted the association of AP2, SP1, and E2F6 with the BRD7 promoter, and promoted association of c-Myc with the BRD7 promoter ( Figure 6D). These transfections also inhibited the activity of the BRD7 promoter ( Figure 6E) and decreased the level of endogenous BRD7 mRNA and protein ( Figure 5B, C). Ectopic overexpression of LRRC4 in U251 cells ( Figure 5D) or endogenous re-expression of LRRC4 induced by 5-Aza-dC treatment in U251, SF126, and SF767 cells also promoted c-Myc association with the BRD7 promoter but disrupted the association of AP2, SP1, and E2F6 ( Figure S6). Treatment with PD98059 or LY294002 alone, or the two in combination, led to a decrease in AP2, SP1, and E2F6 associating with the BRD7 promoter and promoted the association of c-Myc with the promoter. Moreover, these kinase inhibitor treatments also reversed the miR-182-and miR-38-induced BRD7 promoter associations of AP2, SP1, E2F6, and c-Myc ( Figure 6F). These results indicated that miR-182 and miR-381 silencing interfered with or promoted the binding of various transcription factors to the BRD7 promoter. siRNA-mediated silencing of AP2, SP1, or E2F6, and overexpression of c-Myc in U251 cells resulted in downregulation of BRD7 at both the mRNA and protein levels ( Figure 6G).
Discussion miRNAs can also function as tumor suppressors or oncogenes to induce cellular transformation and tumorigenesis [23].The facts that miRNAs can regulate the expression of specific genes, including those that are differentially expressed in malignant cells, and that they themselves are differentially expressed in the malignant state, have made miRNAs attractive therapeutic targets [24]. LRRC4 has been characterized as a tumor suppressor gene involved in glioma formation [4], and our previous study indicated that LRRC4 is a target of miR-381 [17]. In this study, we demonstrated that LRRC4 is also a target of miR-182. miR-182 and miR-381 were also determined to be involved in the pathological progression of astrocytoma by targeting LRRC4. Suppressing endogenous expression of miR-182 and miR-381, respectively, restored the activation of LRRC4 in gliomas, and inhibited glioma cell proliferation in vitro and growth of subcutaneously transplanted tumor in vivo. Furthermore, it inhibited the growth of intracranial transplanted tumors, presumably by surpassing the blood-brain barrier. Thus, miR-182 and miR-381 may represent useful therapeutic targets for treatment of these tumors.
The precise roles of miR-182 and miR-381 in relation to LRRC4 expression in gliomas were investigated by the miRNA silencing tool of locked nucleic acids. LNAs are a class of bicyclic conformational analogs of RNA, which exhibit high binding affinity to their complementary RNA molecules and are highly stable in vivo [25]. In the current study, LNA-mediated miR-182 and miR-381 silencing was applied and found to restore the expression of LRRC4 in gliomas. We found that LNA-mediated miR-182 and -381 silencing decreased the expression of BRD7, which was striking in its similarity to the results obtained with LRRC4 overexpression. Thus, it was suggested that LNA-mediated miR-182 and miR-381 silencing down-regulated the expression of BRD7 by restoring LRRC4 expression in gliomas. BRD7 has been characterized as a potential nuclear transcription factor that regulates cell cycle progression through the Rb/E2F pathways [26], and has been shown to induce cell apoptosis in NPC [27].The study presented herein demonstrated that LNA-mediated miR-182 and miR-381 silencing in gliomas blocked cell cycle progression in the G0/G1 phase by regulating pRb and E2F3 and inhibited cell proliferation in vitro and growth in vivo. It was also found that LNA-mediated miR-182 and miR-381 silencing induced marked differentiation of tumor cells towards a noncancerous status.
In previous researches, we had focused on the effect of LRRC4 on the ERK/MAPK and PI-3K/AKT signaling pathways in gliomas [4,7], and transcriptional regulation of BRD7 expression in NPCs [18]. ERK/MAPK and PI-3K/AKT signaling are major cell survival pathways, and play a key role in diverse physiological and pathological processes [25]. In the present study, we demonstrated that LNA-mediated miR-182 and miR-381 silencing can affect the expression and activity of transcription factors that have binding sites in the BRD7 promoter, including AP2, SP1, E2F6, and c-Myc. Moreover, these particular factors are known to be regulated by LRRC4 through the K-Ras/c-Raf/ERK/MAPK and PI-3K/AKT signaling pathways. miR-182 and miR-381 silencing decreased the expression and activity of AP2, SP1, and E2F6, but increased the expression and activity of c-Myc. Silencing also inhibited the promoter activity and expression of BRD7, as did knock-down of AP2, SP1, and E2F6, and overexpression of c-Myc. These results indirectly confirmed that c-Myc is a negative regulator of BRD7, and that AP2, SP1, and E2F6 are positive regulators. Of course, this indirect effect may also be explained by a more complex molecular mechanism involving other factors, and we plan to investigate such potential interactions in future studies.

Cell Culture
Three human glioma-derived cell lines, U251, SF126 and SF767, and the rat C6 glioma cell line were obtained from the Cell Center of Peking Union Medical College in China, and no Figure 2. miRNA-182 and miR-381 or BRD7 expression is inversely related to LRRC4 expression in gliomas. (A) miRNA-182 and miR-381 or BRD7 expression is inversely related to LRRC4 expression in glioma tissues. miR-182 and miR-381 expression levels were assessed by ISH. IHC was used to detect the protein expression of LRRC4 and BRD7. (B) miRNA-182 and miR-381 or BRD7 expression is inversely related to LRRC4 expression in normal brain and WHO grade I, II, III astrocytomas, and grade IV glioblastoma. (C) The total gray value of (B). Image analysis and total gray value were estimated by the GSM-2000P pathology image analysis system. doi:10.1371/journal.pone.0084146.g002 miR-182 and miR-381 Are Targets for Glioma Therapy PLOS ONE | www.plosone.org authentication was performed by the authors since the Cell Center is reputed to be a reliable source with carefully monitored laboratory practices that adhere to international standards. U251 cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) with 10% fetal calf serum (FCS) and standard antibiotics. SF126, SF767 and C6 cells were cultured in minimum essential medium (MEM). All cells were maintained at 37uC under an atmosphere of 5% CO 2 and 95% air.

Patient samples
Sixty-seven human primary brain tumor samples were obtained from randomly selected cancer patients (n = 67) treated at Xiangya Hospital, Hunan, China. All the diagnoses were confirmed by pathology. Written informed consent was obtained from each patient participating in the study prior to surgery. All of the protocols were reviewed and approved by the Joint Ethics Committee of the Central South University Health Authority and performed in accordance with national guidelines.

Luciferase assays
The 39-untranslated regions (UTRs) of the LRRC4 or The cells were plated into a 24-well plate and incubated for 24 h before transfection. pMIR-REPORT vectors harboring wildtype (WT) or mutant 39-UTR LRRC4 sequences were cotransfected into cells along with the miR-182 or miR-381 constructs using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). Luciferase assays were performed 24-48 h after transfection by using the Luciferase Reporter Assay System (Promega, Shanghai, China).

Real-time quantitative (q)RT-PCR analysis
The total RNAs were extracted from cells or tissues with TRIzol reagent (Invitrogen, Wuhan, China). Reverse transcription reactions were performed with reagents from a SYBR-green-containing PCR kit (GenePharma, Shanghai, China). The primers for qRT-PCR detection of miRNA were designed based on the miRNA sequences provided by the Sanger Center miRNA Registry and were synthesized and purified by Shanghai GenePharma. Human U6 small nuclear (sn)RNA was used for normalization. Primers used are as follows: miR-182, 59- In situ hybridization (ISH), immunochemistry (IHC) and immunofluorescence analysis miR-182 or miR-381 miRCURY TM LNA custom detection probes (Exiqon, Vedbaek, Denmark) were used for ISH. The 59-39 sequence (enhanced with LNA) was TCAGGAACTGCC-TTTCTCTCCA or ACAGAGAGCTTGCCCTTGTATA with a DIG label at both the 59 and 39 ends. Hybridization, washing, and scanning were carried out according to the manufacturer's instructions. IHC studies were performed using the standard streptavidin/peroxidase staining method as described previously [17]. Image analysis and total gray value estimation were conducted by the GSM-2000P pathology image analysis system (Heima, Zhuhai, China). Immunofluorescence staining was performed as previously described [17]. Stained cells were viewed with a Zeiss LSM510 laser scanning fluorescence microscope. 5-Aza-29-deoxycytidine (5-Aza-dC) demethylation treatment U251, SF126, and SF767 cells were grown for 48 h in the presence of 10 mM 5-Aza-dC (Sigma-Aldrich, St. Louis, MO, USA), as previously described [5]. Fresh 5-Aza-dC was added every 24 h.

BRD7 promoter construct
The sequence of BRD7 promoter was previously described in [18].

Electrophoretic mobility shift assays (EMSA)
Complementary oligonucleotides derived from the human BRD7 promoter regions and containing putative Sp1, E2F6, c-Myc and E2F-binding sites, as previously described [19], were synthesized by Invitrogen. Nuclear extracts were isolated from cells using the Nuclear and Cytoplasmic Extraction Reagent (Pierce Chemical, Rockford, IL, USA). EMSAs with nuclear extracts and biotin-labeled BRD7 promoter oligonucleotides were performed according to the LightShift Chemiluminescent EMSA Kit's instructions (Pierce Chemical).

Cell proliferation assay
This procedure was carried out as previously described [4].

Cell cycle assay
This procedure was carried out as previously described [4].

Subcutaneously transplanted tumor formation assay in nude mice
This procedure was carried out as previously described [17].

Intracranial transplanted tumor model in rats
All experiments were carried out with the approval of the Animal Care and Use Committee of Central South University. Sprague-Dawley rats, 4-6 weeks old and weighing 200-250 g were used. For implantation to brain parenchyma, C6 glioma cells were suspended at a concentration of 2610 6 cells in 10 mL PBS. Rats were anesthetized by intraperitoneal injection of ketamine (40 mg/kg) in combination with xylazine (10 mg/kg). Each rat's head was mounted in a stereotactic head holder and a burr hole was made 4 mm from the brain surface, 4 mm lateral and 0.5 mm posterior to the Bregma for determining cerebral blood flow in the subcortex region. A stainless cannula was inserted through the hole into the S1 region to make an injection pocket. Cells were injected via a stereotactic pump at 1 mL/min using a 25 mL syringe. After injection of the cells, the needle was allowed to remain inside the tissue for 5 min to prevent the cell suspension from leaking out. After the syringe was extracted, the hole was sealed with sterile bone wax. Strict sterility was maintained throughout the procedure.

In vivo magnetic resonance imaging
On the 7th day after implantation of C6 glioma cells, rats were randomly divided into four groups (n = 5 per group). Aliquots of 200 mL of 0.9% saline solution containing 2 mg of scrambled, LNA-anti-miR-182, or LNA-anti-miR-381, or 1 mg of LNA-anti-miR-182+1 mg of LNA-anti-miR-381 were administered via intraperitoneal injection for one day. On the 28th day after implantation of the C6 glioma cells, rats were anesthetized as described above and intravenously injected with 0.2 mL/kg gadopentetic acid dimeglumine salt injection (Bayer Schering Pharma, Germany). MRI was carried out for T1WI with FSE sequence and T2WI with FSE sequence, with section thickness of 2 mm and gap of 0.1 mm. Tumor volume was calculated as follows: V = L6l 2 60.5 mm 3 , where L and l represented the larger and smaller tumor diameters, respectively.

Statistical analysis
Differences between groups were tested by the Student's t-test or one-way ANOVA using the SPSS 11.0 program (Chicago, IL, USA). Spearman's correlation test was used to evaluate the pairwise expression correlation between miR-182, miR-381, BRD7 and LRRC4 in gliomas. A p-value of less than 0.05 was considered statistically significant.