Gliomas are the most common brain tumors of the adult central nervous system. One of the insidious biological features of glioma is the potential of single cells to invade normal brain tissue. Invasive tumor cells escape surgical removal and radiation exposure. The mechanism of invasion is quite complex and involves integrated biochemical processes requiring a coordinated effort managing a number of intracellular and extracellular interactions.1 Tumor cells achieve this by excessive production of several proteases and by modifying ECM, creating local access through surrounding tissue and migrating into other parts of the brain.2, 3 Also, members of the Ras superfamily of GTPases, most notably the Rho proteins, plays a prominent role in cell migration.4
MicroRNAs (miRNAs), which encode small non-coding RNAs of approximately 22 nucleotides, are now recognized as a very large gene family present throughout the genomes of plants and metazoans.5–7 Evolutionally conserved miRNAs have stem-loop structures and are often found as clustered sequences. Mature miRNAs processed by two-step cleavage involving Drosha and Dicer are thought to regulate the negative expression of a large number of genes carrying target sites within 3' untranslated regions, whereas recent evidence points to miRNA utilization of multiple mechanisms for gene silencing. Recent work has revealed important roles of miRNAs and miRNA processing in tumorigenesis.8, 9 A large set of miRNAs is overexpressed in human tumors compared to normal tissues,10, 11 and gene silencing by miRNAs enhances tumor cell growth.12 These findings suggest that this class of regulators includes enhancers of tumor progression.
The miR-10b gene is located in the middle of the HOXD cluster on chromosome 2q31, near HOXD4.13 A combination of mouse and human cells exhibited that miR-10b is highly expressed in metastatic breast cancer cells and positively regulates cell migration and invasion.13 Overexpression of miR-10b in otherwise nonmetastatic breast tumors initiates robust invasion and metastasis. The miR-10b proceeds to inhibit translation of HOXD10, resulting in increased expression of a well-characterized prometastatic gene, RhoC. In addition, the level of miR-10b expression in primary breast carcinomas correlates with clinical progression. Recently, miR-10b was found to be overexpressed in hepatocellular carcinoma compared to benign hepatic tumors and nontumor liver tissues.14 These findings suggested a novel pathway which regulates tumor cell invasion and migration by miR-10b.
miR-10b was previously reported to be upregulated in glioblastoma tissues.15 Our preliminary miRNA microarray studies also showed miR-10b upregulation in glioblastoma; therefore, we examined the expression level of miR-10b in 43 cases of various gliomas and found that miR-10b was upregulated in all gliomas compared to nontumor brain tissues. Overexpression levels of miR-10b were highly associated with higher grade gliomas. Also, expression levels of invasive factors RhoC and urokinase-type plasminogen activator receptor (uPAR) were associated with expression levels of miR-10b. In addition, high expression levels of miR-10b were associated with multifocal tumors and dissemination. From these findings, it is possible that miR-10b might play some role in the invasion of glioma.
Material and methods
Clinical specimens and cell lines
Glioma tissues were obtained from therapeutic procedures performed as routine clinical management at our institution. Tissue samples were resected during surgery and immediately frozen in liquid nitrogen for subsequent total RNA extraction. A total of 43 gliomas (glioblastoma (GBM) = 17, anaplastic astrocytoma (AA) = 6, low-grade glioma (LGA) = 10 (diffuse astrocytoma = 7, pilocytic astrocytoma = 3), oligodendroglioma (OG) = 6, ependymoma (EP) = 4), 5 meningiomas, 3 schwannomas and 5 non-neoplastic brain specimens were included in our study. Six GBM cell lines (A172, U87MG, T98G, U251, YH13 and SF126) were also used. GBM cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing glutamine, 10% fetal bovine serum and penicillin/streptomycin. Cells were grown at 37°C in a 5% CO2 incubator.
RNA isolation and fluorescent labeling
Total RNA for miRNA microarray was obtained from tissue samples using Trizol reagent (Invitrogen, Carlsbad, CA). To prepare cellular miRNA, small RNA containing miRNA was isolated from total RNA using the RNeasy MinElute Cleanup Kit (QIAGEN Inc., Mississauga, Ontario, Canada), as described previously. The isolated small RNA (∼1 μg) was subjected to direct labeling with a fluorescent dye using the PlatinumBright 647 Infrared nucleic acid labeling kit (KREATECH, Amsterdam, The Netherlands), according to the manufacturer's instructions. After labeling, the RNA was purified from fluorescent-free substrates using KREApure columns (KREATECH) according to the manufacturer's instructions, and used in hybridization. For real-time reverse transcriptase polymerase chain reaction (RT-PCR), total RNA were extracted from frozen tissue samples or cell lines using the mirVana™ miRNA Isolation Kit (Ambion, Austin, TX) according to the manufacturer's instructions.
Hybridization with DNA chips
Hybridization was carried out using Genopal-MICH07 DNA chips or Genopal-MICH DNA chips (Mitsubishi Rayon, Tokyo, Japan), in which 188 or 127 oligonucleotide DNA probes are installed to detect human miRNA in 150 μl hybridization buffer (2× SSC, 0.2% SDS, and ∼1 μg of heat-denatured labeled RNA) at 50°C overnight. After hybridization, the DNA chips were washed twice in 2× SSC containing 0.2% SDS at 50°C for 20 min followed by washing in 2× SSC at 50°C for 10 min, and then hybridization signals were examined and analyzed using a DNA chip image analyzer according to the manufacturer's instructions (Mitsubishi Rayon). Chip analysis was repeated at least twice and hybridized signal intensities were analyzed as described previously. DNA chip data were compared using KURABO custom analysis services (KURABO Industries, Osaka, Japan, the authorized service provider of Mitsubishi Rayon, Tokyo, Japan).
Real-time RT-PCR for miRNAs
Total RNA for real-time RT-PCR was obtained from tissue samples using a mirVana™ miRNA Isolation Kit (Applied Biosystems, Foster City, CA). TaqMan microRNA assays (Applied Biosystems) were used to quantify mature miRNA expressions, as previously described. RNU6 (Applied Biosystems) was used as the endogenous control for miRNA expression studies. For miRNA expression quantification, each reverse-transcriptase (RT) reaction consisted of 50 ng of purified total RNA, 1× RT buffer, dNTPs (each at 0.375mM), 5 U μl−1 MultiScribe reverse transcriptase, 50 nM stem-loop RT primer and 0.38 U μl−1 RNase inhibitor (Applied Biosystems). RT reactions were incubated at 16°C for 30 min, 42°C for 30 min and 85°C for 5 min. Real-time PCR reactions for miRNA from cells were performed in quadruplicate in 20 μl volumes. The real-time reaction mix consisted of 1.33 μl RT product, 1 μl of 20× TaqMan microRNA assay mix and 10 μl TaqMan 2× universal PCR Master Mix. Quantitative miRNA expression data were acquired and analyzed using an Applied Biosystems 7500 real-time PCR system (Applied Biosystems).
Real-time RT-PCR for RhoC and uPAR
Total RNA was obtained from tissue samples using a mirVana™ miRNA Isolation Kit (Applied Biosystems), which is also used for real-time RT-PCR in miRNA studies. Reverse transcription was performed using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems), according to the manufacturer's instructions. Expression levels of mRNA of RhoC and uPAR were assayed quantitatively by real-time RT-PCR using TaqMan® Gene Expression Assays (Applied Biosystems). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the endogenous control. Quantitative mRNA expression data were acquired and analyzed by ΔΔ-Ct method using an Applied Biosystems 7500 real-time PCR system (Applied Biosystems).
Archived paraffin blocks of glioma patients in the Department of Pathology of our hospital were used for immunostaining. These cases were reviewed by experienced pathologists. The serial sections were deparaffinized, immersed in methanol with 0.3% hydrogen peroxide and heated in 0.01 M citrate buffer (pH 6.0; for RhoC) or 1 mM ethylenediamine tetraacetic acid (EDTA) buffer (pH 9.0; for uPAR) for 15 min by autoclaving (121°C, 2 atm). The sections were then incubated with primary antibodies at 4°C overnight. The applied antibodies were goat anti-uPAR polyclonal antibody and goat anti-RhoC polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) The sections were allowed to react with peroxidase-conjugated anti-goat IgG polyclonal antibody (Histofine Simple Stain MAX-PO; Nichirei, Tokyo, Japan) for 45 min, and reaction products were visualized by immersing the sections in 0.03% diaminobenzidine solution containing 2 mM hydrogen peroxide for 1–5 min. Nuclei were lightly stained with Mayer's hematoxylin. For control studies of the antibodies, serial sections were treated with phosphate-buffered saline, normal goat IgG monoclonal antibody (Nichirei; subclass IgG2a) instead of the primary antibodies, and were confirmed to be unstained. Immunohistochemical reactivity for uPAR and RhoC was evaluated and classified into 4 groups: (−) negative in neoplastic cells, (+) weakly positive in neoplastic cells, (++) moderately positive in neoplastic cells and (+++) strongly positive in neoplastic cells.
Differences in miR-10b expression levels between 2 groups were analyzed using the Mann–Whitney U test. The correlation between miR-10b and mRNA levels of RhoC and uPAR was assessed using Spearman's rank tests. Differences between 2 groups for immunohistochemistry were analyzed using the Mann–Whitney U test. P < 0.05 was considered significant.
Expression of miRNAs in GBM
For miRNA profiling in glioma, we first isolated and compared miRNA expression profiles from brain and glioma tissue. Using this method, we analyzed tissue samples from 3 GBM patients. To yield a specific case by case matching pair of tumor and control samples, we obtained samples from the centre of the tumor and from a peripheral glial brain area. We confirmed the histology of each sample by H&E staining; many GBM cells were observed in specimens from the tumor centre, while peripheral glial brain specimens did not show any evidence of distinct GBM cells (Supporting Information Figure 1). Six microarray hybridization studies were performed on 3 different pairs of tumor and brain-derived RNA. To analyze different miRNA expressions in GBM, the ratio of each tumor sample and the corresponding normal brain sample was calculated. The miRNAs that were upregulated differentially in GBM included miR-10b, miR-21, miR-183, miR-92b, miR-106b (Table I and Supporting Information Figure 2). On the other hand, miRNAs that were downregulated in glioma included miR-302c*, miR-379, miR-329, miR-134, miR-369-3p, miR-221 (Table I and Supporting Information Figure 2). Among upregulated miRNAs, the expression of miR-10b was higher in tumor than peripheral brain tissue, and was in excess of 1.53-, 6.67- and 90.3-fold (mean 32.8 ± 28.80). Another consistent feature of the miRNA expression pattern was the striking upregulation of miR-21 in all three gliomas compared to its expression in the control (1.94-, 2.87- and 5.16-fold; mean 3.32 ± 0.95). miR-21 is known to be an antiapoptotic factor in human GBM cells.16
Table I. Microarray Analysis of miRNA Expression in Human Glioblastoma
Up regulated (>1.5)
Down regulated (<0.5)
miR-10b is upregulated in GBM cell lines
We focused on the expression of miR-10b because it was the most overexpressed miRNA in GBM in our microarray study, and a previous report has also shown the upregulation of miR-10b in GBM.15 First, we examined miR-10b expression in GBM cell lines. We performed real-time RT-PCR on a panel of 6 human GBM cell lines (A172, T98G, U87MG, U251, YH13 and SF126) and 5 nontumor brain tissues, which were used as a control. And miR-10b expression of each cell line was compared to the average expression level of miR-10b of 5 nontumor brain tissues. As shown in Figure 1, miR-10b levels of all cell lines were higher than that of nontumoral brain tissue. miR-10b expression in A172, U87 and T98G cells was relatively high. In contrast, expression levels of miR-10b in U251, YH13 and SF126 cells were relatively low.
miR-10b is upregulated in glioma
Next, we analyzed miR-10b expression levels in primary tumor samples from 43 glioma patients (GBM 17; AA 6; LGA 10; OG 6; EP 4) and 6 non-neoplastic brain tissues. When compared to nontumoral brain tissue, miR-10b expression levels in all glioma samples were high (Fig. 2a). In glioma samples, the fold difference of miR-10b expression levels ranged from 0.08- to 364-fold, with a mean of 46.7-fold. When expression levels of miR-10b were compared between subgroups, miR-10b was extremely overexpressed in GBMs. Note that miR-10b expression was correlated with the tumor grade and malignancy in astrocytic tumors. Mann–Whitney U test analysis showed that the mean of miR-10b levels in GBM was significantly higher that that in LGA and OG, as shown in Figure 2b (p < 0.01, p = 0.015, respectively). No statistically significant difference in miR-10b expression was observed between GBM and AA. Also, the mean miR-10b level in AA was significantly higher that that in LGA (p = 0.039). The mean miR-10b level in high-grade glioma, including GBM and AA, was statistically significantly higher than in low-grade gliomas, including LGA, OG and EP (p < 0.01, Fig. 2c). miR-10b expression levels of meningiomas and schwannomas were low compared to gliomas (mean of 2.0-fold for meningioma and 4.8-fold for schwannoma) (Fig. 2b).
miR-10b expression correlates with mRNA expression of RhoC and uPAR
Ma et al. have shown that miR-10b inhibits the translation of HOXD10 mRNA, thereby affecting the expression of downstream targets of this transcription factor. Two effectors of HOXD10 are RhoC and uPAR, which functions in invasion, migration and metastasis13; therefore, we next examined the mRNA expressions of RhoC and uPAR in the same glioma samples using real-time RT-PCR. As shown in Figure 3a and 3b, there were significant correlations between miR-10b and mRNA expressions of RhoC (p < 0.001, Fig. 3a), and between miR-10b and mRNA expressions of uPAR (p = 0.006, Fig. 3b) in all glioma samples. When analyzing the correlation in only high-grade gliomas, including GBM and AA, there were significant correlations between the expressions of miR-10b and RhoC (p < 0.001), and miR-10b and uPAR (p = 0.001) (data not shown).
Correlation between miR-10b with protein expression of RhoC and uPAR
To examine the correlation between miR-10b and the protein expressions of RhoC and uPAR, immunohistochemical analysis was performed. Immunohistochemical reactivity was evaluated and classified into 4 groups: negative (−), weak (+), moderate (++), and strong (+++) (Fig. 4). All tumor samples were obtained from the center of the tumor. Also, all glioma samples were classified into 2 groups based on the levels of miR-10b: over 25-fold relative expression level of miR-10b was defined as the high miR-10b expression group. In 16 cases in the high miR-10b expression group, moderate and strong expressions of RhoC protein were observed in 4 (25%) and 2 (13%) cases, respectively (Table II). On the other hand, the low miR-10b expression group was seen in 2 (7%) and 0 (0%) cases, respectively. RhoC protein expression in the miR-10b high expression group was significantly higher than that in the low expression group (p = 0.009, Mann–Whitney's U-test) Also, in 16 high miR-10b expression samples, moderate and strong expressions of uPAR were found in 4 (25%) and 1 (6%) cases, respectively, whereas in the low miR-10b expression group, there were 4 (15%) and 0 (0%) cases, respectively (Table II). There was a statistically significant difference between high and low miR-10b expression groups in uPAR immunostaining (p = 0.014, Mann–Whitney's U-test).
Table II. Correlation Between miR-10b and Immunostaining of RhoC and uPAR
Interpretation RhoC and uPAR immunostaining is described in “Material and methods” immunostaining was graded as negative (−), weak (+), moderate (++), and strong (+++).
: statistically significant, Mann-Whitney's U test.
miR-10b expression correlates with multifocal lesion of malignant glioma
To analyze the association of miR-10b expression with glioma invasion, we examined MR images of lesions in GBMs and AAs and classified them into two groups: multifocal Gd-enhanced lesions and a single lesion (Fig. 5b). There were 7 multifocal lesions of the 21 cases of GBM and AA (GBM, 5 cases; AA, 2 cases). In the multifocal group, the relative level of miR-10b ranged from 24.5- to 364-fold, with a mean of 157.9-fold. On the other hand, in the single group, the relative level of miR-10b ranged from 0.3- to 271.5-fold, with a mean of 56.7-fold. There was a statistically significant difference between the multifocal and single groups (p = 0.02) (Fig. 5a).
Aberrant patterns of miRNA expression have already been described in GBM.15–18 In our study, we examined the expression levels of miR-10b in various gliomas, and found that miR-10b was upregulated in not only high-grade gliomas but also low-grade astrocytomas, OGs and EPs. Moreover, expression levels of miR-10b are associated with glioma malignancy, and are much higher in GBM than in other gliomas. In addition, mRNA expressions of RhoC and uPAR, which are known to contribute to glioma invasion and migration, were correlated with miR-10b expression in gliomas. Furthermore, miR-10b expression in high-grade glioma with multifocal lesions on enhanced MRI was statistically significantly higher than that with a single lesion. These findings indicated that miR-10b might play some role in the invasion and migration of glioma cells.
miRNAs expression in glioma
Thus far, 2 reports have studied miRNA expression profiling in glioma.15, 16 The first report by Chen showed the overexpression of miR-21, -135, -138, -291-5' and -347 in GBM, but no overexpression of miR-10b.16 Ciafre and colleagues then described extensive modulation of a set of miRNAs in primary GBMs, in which miR-10b, -130a, -221, -125, -9-2, -21, -25 and -123 were shown to be upregulated compared to nontumorous brain tissues.15 Among these, the most upregulated miRNA was miR-10b. They also showed the overexpression of miR-10b in GBM tissues by Northern blot analysis. In our microarray analysis, miR-10b, -21, -183, -92b and -106b were upregulated compared to non-neoplastic brain controls. In a real-time RT-PCR study, miR-10b was upregulated in all gliomas; in particular, GBM expressed a high level of miR-10b. In our study, miR-10b was overexpressed not only in glioma tissues but also in glioma cell lines. A previous study, however, showed no significant alteration of the expression levels of miR-10b in glioma cell lines. There are several reasons for this discrepancy. One is the analysis method; we used quantitative real-time RT-PCR and the previous study used Northern blot analysis. Another is thought to be the cell culture conditions, such as the culture medium and cell confluence.
Our study showed that expression levels of miR-10b were associated with glioma grading and malignancy. Several recent studies have shown that miR-10b expression is associated with tumor malignancy. miR-10b was found to be overexpressed in hepatocellular carcinoma when compared to benign hepatic tumors or nontumor liver tissues.14 miR-10b was also upregulated in acute myeloid leukemia (AML) cells with mutation of the nucleophosmin (NPM1) gene.19 Ma et al. identified miR-10b as an miRNA highly expressed in metastatic breast tumors that promotes cell migration and invasion.13 They showed that miR-10b inhibits the translation of mRNA encoding homeobox D10 (HOXD10), which represses the expression of genes involved in cell migration and extracellular matrix remodeling.
High-grade glioma cells have high invasive potential, and multifocal lesions and dissemination are frequently observed; however, not all high-grade gliomas show high invasive potential. In our study, there were 7 high-grade gliomas with multifocal lesions, whose expression levels of miR-10b were statistically significantly higher than that with a single lesion, a very interesting finding which indicates that miR-10b could have some role in tumor cell invasion and migration in glioma. Also, because miR-10b expression of glioma cell lines A172, U87 and T98 were relatively high and U251, YH13 and SF126 were low in our study, it is interesting to investigate the association between miR-10b expression and the invasive potential or migration levels using these cell lines.
miR-10b and regulating molecules
miR-10b is induced by TWIST,13 which is a master regulator of morphogenesis and plays an essential role in tumor metastasis.20 Interestingly, TWIST was detected in a large majority of human glioma and increased TWIST mRNA levels were associated with the highest grade gliomas.21 This report supports our findings of an miR-10b expression pattern in glioma. miR-10b inhibits the translation of mRNA encoding HOXD10, which modulates many genes that promote invasion, migration, extracellular matrix remodeling and tumor progression, including uPAR, RhoC, α3 integrin, β integrin and matrix metalloprotease-14 (MMP-14).22 The β1 subunit of integrin plays an important role in glioma biology,23 and α3β1 integrin has been shown to be a key regulator of glioma cell migration.24 uPA is a specific serine protease, and uPA binds to its specific receptor, uPAR, directing plasmin activity to the migrating tumor cell surface. High expressions of uPA and uPAR correlate with an invasive phenotype of glioma cells,25 and uPAR has been shown to be overexpressed in GBM.26
Rho family proteins are regulators of extracellular stimuli-mediated signaling pathways that control numerous intracellular processes, including actin cytoskeletal organization, gene expression, invasion and cell cycle progression. Two GTPases, Rho and Rac, are necessary to enable the dynamic process of cell detachment from the ECM and their reattachment to it, and are therefore essential for cell migration. RhoC has been identified as an especially important player in metastasis and its expression correlates with the metastatic spread of various types of carcinomas.27, 28 In glioma, however, RhoC expression remains poorly understood. RhoC mRNA was upregulated in oligodendroglioma compared to normal brain tissue.29 Also, Zhai reported that radiation enhances the invasive potential of primary GBM cells via activation of the Rho signaling pathway.30
In our study, expression of uPAR and RhoC correlated with expression levels of miR-10b. These results indicate that miR-10b might have some role in invasion in gliomas.
Regulation of migration and invasion by miRNAs
Recently, several microRNAs were identified as either promoters or suppressors of invasion and metastasis. miR-21, which is frequently overexpressed in GBM, regulates invasion and metastasis by targeting metastasis-related tumor suppressor genes, such as TPM1 and maspin.31 Also, miR-373 and miR-520c stimulate cancer cell migration by suppressing CD44.32 As CD44 is related to glioma invasion,1 it may be interesting to examine the expression of these miRNAs in glioma. In contrast, miR-335 suppresses metastasis and migration by targeting the progenitor cell transcription factor SOX4 and extracellular matrix component tenascin C.33 Whether other miRNA family members can also stimulate or suppress tumor migration and invasion remains to be investigated.
miRNA studies in tumor cell invasion and migration have just started. From our studies, it is possible that miR-10b might play some role in the invasion of glioma.
This work was supported in part by a Grant-in-Aid for Scientific Research to Eiji Kohmura (15659337), Katsu Mizukawa (19790998), and Takashi Sasayama (17790968) from the Japanese Ministry of Education, Culture, Sports, Science and Technology. The authors also thank Miss Ayumi Katoh for technical assistance with RT-PCR and immunohistochemical analysis.